fix(tests): disable telemetry in afterEach to prevent cross-file fetch leak

[la14-1]

Why: 2 tests fail on every bun test run (hetzner-cov, digitalocean-token) because the telemetry test enables _enabled = true globally and never resets it — every subsequent logWarn/logError triggers fire-and-forget fetch() calls that corrupt other tests' callCount-based mocks.

Root cause

telemetry.test.ts deletes BUN_ENV/NODE_ENV and calls initTelemetry() to enable telemetry for testing. Once _enabled is true, it stays true for the entire process. Any logWarn() or logError() in concurrent test files (hetzner, digitalocean) fires sendEvent() → fetch() through the calling test's mock, consuming callCounts.

Fix (2 lines of real change)

1. Export _testHelpers.enabled getter/setter from telemetry.ts
2. Set telemetryTestHelpers.enabled = false in telemetry.test.tsafterEach

This fixes the leak at the source rather than patching each affected test to filter PostHog URLs.

Relationship to #3341 and #3353

Both existing PRs work around the symptom (patching individual test mocks to ignore telemetry URLs). This PR fixes the root cause — once merged, those workarounds become unnecessary and no future test can be affected by telemetry fetch leaks.

Test plan

• bun test passes with 0 failures (2108 pass, 5376 expect() calls)
• Verified on 2 consecutive runs — no flakiness
• biome check passes on changed files

-- spawn-refactor/test-engineer

[la14-1]

Overlap note: This PR fixes the root cause (telemetry _enabled leaking across test files) of the same 2 flaky tests addressed by #3353 and #3365.

• fix(tests): disable telemetry in afterEach to prevent cross-file fetch leak #3358 (this PR) plugs the leak at the source via afterEach reset
• fix(tests): filter PostHog telemetry from fetch mocks #3353 adds URL filtering in affected tests (simpler defensive fix)
• fix(test): resolve 2 flaky tests caused by parallel fetch mock interference #3365 uses URL-based routing with per-domain counters (more robust defensive fix)

Recommended merge order: this PR first (root cause), then optionally #3365 (defense in depth). #3353 can likely be closed if both land.

All CI checks passing. No merge conflicts.

-- refactor/pr-maintainer

[la14-1]

This PR has been open for 3+ days with all CI checks passing (ShellCheck, Biome Lint, Unit Tests, Mock Tests, macOS Compatibility — all SUCCESS). It is mergeable with no conflicts.

This is the root-cause fix for the flaky test issue (telemetry leak). Recommended merge order: this PR first, then #3365 (defense in depth). #3353 can be closed after both land.

Requesting human review when convenient.

-- spawn-refactor/pr-maintainer

[la14-1]

Dedup update:#3376 (opened Apr 30) supersedes this PR — it also modifies telemetry.ts with a root-cause fix (runtime env-var guard in sendEvent()) plus adds SPAWN_TELEMETRY=0 in preload.ts for defense-in-depth. Both PRs fix the same leak but #3376 is more comprehensive and will conflict with this one on telemetry.ts.

-- refactor/pr-maintainer

[la14-1]

Closing as superseded by #3376, which provides a more comprehensive fix for the same telemetry singleton leak (runtime env-var guard in sendEvent() + SPAWN_TELEMETRY=0 in test preload).

-- refactor/pr-maintainer

[la14-1]

Superseded by #3376.

[la14-1]

Why: 2 tests fail on every bun test run (hetzner-cov, digitalocean-token) because the telemetry test enables _enabled = true globally and never resets it — every subsequent logWarn/logError triggers fire-and-forget fetch() calls that corrupt other tests' callCount-based mocks.

Root cause

telemetry.test.ts deletes BUN_ENV/NODE_ENV and calls initTelemetry() to enable telemetry for testing. Once _enabled is true, it stays true for the entire process. Any logWarn() or logError() in concurrent test files (hetzner, digitalocean) fires sendEvent() → fetch() through the calling test's mock, consuming callCounts.

Fix (2 lines of real change)

1. Export _testHelpers.enabled getter/setter from telemetry.ts
2. Set telemetryTestHelpers.enabled = false in telemetry.test.tsafterEach

This fixes the leak at the source rather than patching each affected test to filter PostHog URLs.

Relationship to #3341 and #3353

Both existing PRs work around the symptom (patching individual test mocks to ignore telemetry URLs). This PR fixes the root cause — once merged, those workarounds become unnecessary and no future test can be affected by telemetry fetch leaks.

Test plan

• bun test passes with 0 failures (2108 pass, 5376 expect() calls)
• Verified on 2 consecutive runs — no flakiness
• biome check passes on changed files

-- spawn-refactor/test-engineer

[la14-1]

Overlap note: This PR fixes the root cause (telemetry _enabled leaking across test files) of the same 2 flaky tests addressed by #3353 and #3365.

• fix(tests): disable telemetry in afterEach to prevent cross-file fetch leak #3358 (this PR) plugs the leak at the source via afterEach reset
• fix(tests): filter PostHog telemetry from fetch mocks #3353 adds URL filtering in affected tests (simpler defensive fix)
• fix(test): resolve 2 flaky tests caused by parallel fetch mock interference #3365 uses URL-based routing with per-domain counters (more robust defensive fix)

Recommended merge order: this PR first (root cause), then optionally #3365 (defense in depth). #3353 can likely be closed if both land.

All CI checks passing. No merge conflicts.

-- refactor/pr-maintainer

[la14-1]

This PR has been open for 3+ days with all CI checks passing (ShellCheck, Biome Lint, Unit Tests, Mock Tests, macOS Compatibility — all SUCCESS). It is mergeable with no conflicts.

This is the root-cause fix for the flaky test issue (telemetry leak). Recommended merge order: this PR first, then #3365 (defense in depth). #3353 can be closed after both land.

Requesting human review when convenient.

-- spawn-refactor/pr-maintainer

[la14-1]

Dedup update:#3376 (opened Apr 30) supersedes this PR — it also modifies telemetry.ts with a root-cause fix (runtime env-var guard in sendEvent()) plus adds SPAWN_TELEMETRY=0 in preload.ts for defense-in-depth. Both PRs fix the same leak but #3376 is more comprehensive and will conflict with this one on telemetry.ts.

-- refactor/pr-maintainer

[la14-1]

Closing as superseded by #3376, which provides a more comprehensive fix for the same telemetry singleton leak (runtime env-var guard in sendEvent() + SPAWN_TELEMETRY=0 in test preload).

-- refactor/pr-maintainer

[la14-1]

Superseded by #3376.

[la14-1]

Why: 2 tests fail on every bun test run (hetzner-cov, digitalocean-token) because the telemetry test enables _enabled = true globally and never resets it — every subsequent logWarn/logError triggers fire-and-forget fetch() calls that corrupt other tests' callCount-based mocks.

Root cause

telemetry.test.ts deletes BUN_ENV/NODE_ENV and calls initTelemetry() to enable telemetry for testing. Once _enabled is true, it stays true for the entire process. Any logWarn() or logError() in concurrent test files (hetzner, digitalocean) fires sendEvent() → fetch() through the calling test's mock, consuming callCounts.

Fix (2 lines of real change)

1. Export _testHelpers.enabled getter/setter from telemetry.ts
2. Set telemetryTestHelpers.enabled = false in telemetry.test.tsafterEach

This fixes the leak at the source rather than patching each affected test to filter PostHog URLs.

Relationship to #3341 and #3353

Both existing PRs work around the symptom (patching individual test mocks to ignore telemetry URLs). This PR fixes the root cause — once merged, those workarounds become unnecessary and no future test can be affected by telemetry fetch leaks.

Test plan

• bun test passes with 0 failures (2108 pass, 5376 expect() calls)
• Verified on 2 consecutive runs — no flakiness
• biome check passes on changed files

-- spawn-refactor/test-engineer

[la14-1]

Overlap note: This PR fixes the root cause (telemetry _enabled leaking across test files) of the same 2 flaky tests addressed by #3353 and #3365.

• fix(tests): disable telemetry in afterEach to prevent cross-file fetch leak #3358 (this PR) plugs the leak at the source via afterEach reset
• fix(tests): filter PostHog telemetry from fetch mocks #3353 adds URL filtering in affected tests (simpler defensive fix)
• fix(test): resolve 2 flaky tests caused by parallel fetch mock interference #3365 uses URL-based routing with per-domain counters (more robust defensive fix)

Recommended merge order: this PR first (root cause), then optionally #3365 (defense in depth). #3353 can likely be closed if both land.

All CI checks passing. No merge conflicts.

-- refactor/pr-maintainer

[la14-1]

This PR has been open for 3+ days with all CI checks passing (ShellCheck, Biome Lint, Unit Tests, Mock Tests, macOS Compatibility — all SUCCESS). It is mergeable with no conflicts.

This is the root-cause fix for the flaky test issue (telemetry leak). Recommended merge order: this PR first, then #3365 (defense in depth). #3353 can be closed after both land.

Requesting human review when convenient.

-- spawn-refactor/pr-maintainer

[la14-1]

Dedup update:#3376 (opened Apr 30) supersedes this PR — it also modifies telemetry.ts with a root-cause fix (runtime env-var guard in sendEvent()) plus adds SPAWN_TELEMETRY=0 in preload.ts for defense-in-depth. Both PRs fix the same leak but #3376 is more comprehensive and will conflict with this one on telemetry.ts.

-- refactor/pr-maintainer

[la14-1]

Closing as superseded by #3376, which provides a more comprehensive fix for the same telemetry singleton leak (runtime env-var guard in sendEvent() + SPAWN_TELEMETRY=0 in test preload).

-- refactor/pr-maintainer

[la14-1]

Superseded by #3376.

[la14-1]

Why: 2 tests fail on every bun test run (hetzner-cov, digitalocean-token) because the telemetry test enables _enabled = true globally and never resets it — every subsequent logWarn/logError triggers fire-and-forget fetch() calls that corrupt other tests' callCount-based mocks.

Root cause

telemetry.test.ts deletes BUN_ENV/NODE_ENV and calls initTelemetry() to enable telemetry for testing. Once _enabled is true, it stays true for the entire process. Any logWarn() or logError() in concurrent test files (hetzner, digitalocean) fires sendEvent() → fetch() through the calling test's mock, consuming callCounts.

Fix (2 lines of real change)

1. Export _testHelpers.enabled getter/setter from telemetry.ts
2. Set telemetryTestHelpers.enabled = false in telemetry.test.tsafterEach

This fixes the leak at the source rather than patching each affected test to filter PostHog URLs.

Relationship to #3341 and #3353

Both existing PRs work around the symptom (patching individual test mocks to ignore telemetry URLs). This PR fixes the root cause — once merged, those workarounds become unnecessary and no future test can be affected by telemetry fetch leaks.

Test plan

• bun test passes with 0 failures (2108 pass, 5376 expect() calls)
• Verified on 2 consecutive runs — no flakiness
• biome check passes on changed files

-- spawn-refactor/test-engineer

[la14-1]

Overlap note: This PR fixes the root cause (telemetry _enabled leaking across test files) of the same 2 flaky tests addressed by #3353 and #3365.

• fix(tests): disable telemetry in afterEach to prevent cross-file fetch leak #3358 (this PR) plugs the leak at the source via afterEach reset
• fix(tests): filter PostHog telemetry from fetch mocks #3353 adds URL filtering in affected tests (simpler defensive fix)
• fix(test): resolve 2 flaky tests caused by parallel fetch mock interference #3365 uses URL-based routing with per-domain counters (more robust defensive fix)

Recommended merge order: this PR first (root cause), then optionally #3365 (defense in depth). #3353 can likely be closed if both land.

All CI checks passing. No merge conflicts.

-- refactor/pr-maintainer

[la14-1]

This PR has been open for 3+ days with all CI checks passing (ShellCheck, Biome Lint, Unit Tests, Mock Tests, macOS Compatibility — all SUCCESS). It is mergeable with no conflicts.

This is the root-cause fix for the flaky test issue (telemetry leak). Recommended merge order: this PR first, then #3365 (defense in depth). #3353 can be closed after both land.

Requesting human review when convenient.

-- spawn-refactor/pr-maintainer

[la14-1]

Dedup update:#3376 (opened Apr 30) supersedes this PR — it also modifies telemetry.ts with a root-cause fix (runtime env-var guard in sendEvent()) plus adds SPAWN_TELEMETRY=0 in preload.ts for defense-in-depth. Both PRs fix the same leak but #3376 is more comprehensive and will conflict with this one on telemetry.ts.

-- refactor/pr-maintainer

[la14-1]

Closing as superseded by #3376, which provides a more comprehensive fix for the same telemetry singleton leak (runtime env-var guard in sendEvent() + SPAWN_TELEMETRY=0 in test preload).

-- refactor/pr-maintainer

[la14-1]

Superseded by #3376.

[la14-1]

Why: 2 tests fail on every bun test run (hetzner-cov, digitalocean-token) because the telemetry test enables _enabled = true globally and never resets it — every subsequent logWarn/logError triggers fire-and-forget fetch() calls that corrupt other tests' callCount-based mocks.

Root cause

telemetry.test.ts deletes BUN_ENV/NODE_ENV and calls initTelemetry() to enable telemetry for testing. Once _enabled is true, it stays true for the entire process. Any logWarn() or logError() in concurrent test files (hetzner, digitalocean) fires sendEvent() → fetch() through the calling test's mock, consuming callCounts.

Fix (2 lines of real change)

1. Export _testHelpers.enabled getter/setter from telemetry.ts
2. Set telemetryTestHelpers.enabled = false in telemetry.test.tsafterEach

This fixes the leak at the source rather than patching each affected test to filter PostHog URLs.

Relationship to #3341 and #3353

Both existing PRs work around the symptom (patching individual test mocks to ignore telemetry URLs). This PR fixes the root cause — once merged, those workarounds become unnecessary and no future test can be affected by telemetry fetch leaks.

Test plan

• bun test passes with 0 failures (2108 pass, 5376 expect() calls)
• Verified on 2 consecutive runs — no flakiness
• biome check passes on changed files

-- spawn-refactor/test-engineer

[la14-1]

Overlap note: This PR fixes the root cause (telemetry _enabled leaking across test files) of the same 2 flaky tests addressed by #3353 and #3365.

• fix(tests): disable telemetry in afterEach to prevent cross-file fetch leak #3358 (this PR) plugs the leak at the source via afterEach reset
• fix(tests): filter PostHog telemetry from fetch mocks #3353 adds URL filtering in affected tests (simpler defensive fix)
• fix(test): resolve 2 flaky tests caused by parallel fetch mock interference #3365 uses URL-based routing with per-domain counters (more robust defensive fix)

Recommended merge order: this PR first (root cause), then optionally #3365 (defense in depth). #3353 can likely be closed if both land.

All CI checks passing. No merge conflicts.

-- refactor/pr-maintainer

[la14-1]

This PR has been open for 3+ days with all CI checks passing (ShellCheck, Biome Lint, Unit Tests, Mock Tests, macOS Compatibility — all SUCCESS). It is mergeable with no conflicts.

This is the root-cause fix for the flaky test issue (telemetry leak). Recommended merge order: this PR first, then #3365 (defense in depth). #3353 can be closed after both land.

Requesting human review when convenient.

-- spawn-refactor/pr-maintainer

[la14-1]

Dedup update:#3376 (opened Apr 30) supersedes this PR — it also modifies telemetry.ts with a root-cause fix (runtime env-var guard in sendEvent()) plus adds SPAWN_TELEMETRY=0 in preload.ts for defense-in-depth. Both PRs fix the same leak but #3376 is more comprehensive and will conflict with this one on telemetry.ts.

-- refactor/pr-maintainer

[la14-1]

Closing as superseded by #3376, which provides a more comprehensive fix for the same telemetry singleton leak (runtime env-var guard in sendEvent() + SPAWN_TELEMETRY=0 in test preload).

-- refactor/pr-maintainer

[la14-1]

Superseded by #3376.

[la14-1]

Why: 2 tests fail on every bun test run (hetzner-cov, digitalocean-token) because the telemetry test enables _enabled = true globally and never resets it — every subsequent logWarn/logError triggers fire-and-forget fetch() calls that corrupt other tests' callCount-based mocks.

Root cause

telemetry.test.ts deletes BUN_ENV/NODE_ENV and calls initTelemetry() to enable telemetry for testing. Once _enabled is true, it stays true for the entire process. Any logWarn() or logError() in concurrent test files (hetzner, digitalocean) fires sendEvent() → fetch() through the calling test's mock, consuming callCounts.

Fix (2 lines of real change)

1. Export _testHelpers.enabled getter/setter from telemetry.ts
2. Set telemetryTestHelpers.enabled = false in telemetry.test.tsafterEach

This fixes the leak at the source rather than patching each affected test to filter PostHog URLs.

Relationship to #3341 and #3353

Both existing PRs work around the symptom (patching individual test mocks to ignore telemetry URLs). This PR fixes the root cause — once merged, those workarounds become unnecessary and no future test can be affected by telemetry fetch leaks.

Test plan

• bun test passes with 0 failures (2108 pass, 5376 expect() calls)
• Verified on 2 consecutive runs — no flakiness
• biome check passes on changed files

-- spawn-refactor/test-engineer

[la14-1]

Overlap note: This PR fixes the root cause (telemetry _enabled leaking across test files) of the same 2 flaky tests addressed by #3353 and #3365.

• fix(tests): disable telemetry in afterEach to prevent cross-file fetch leak #3358 (this PR) plugs the leak at the source via afterEach reset
• fix(tests): filter PostHog telemetry from fetch mocks #3353 adds URL filtering in affected tests (simpler defensive fix)
• fix(test): resolve 2 flaky tests caused by parallel fetch mock interference #3365 uses URL-based routing with per-domain counters (more robust defensive fix)

Recommended merge order: this PR first (root cause), then optionally #3365 (defense in depth). #3353 can likely be closed if both land.

All CI checks passing. No merge conflicts.

-- refactor/pr-maintainer

[la14-1]

This PR has been open for 3+ days with all CI checks passing (ShellCheck, Biome Lint, Unit Tests, Mock Tests, macOS Compatibility — all SUCCESS). It is mergeable with no conflicts.

This is the root-cause fix for the flaky test issue (telemetry leak). Recommended merge order: this PR first, then #3365 (defense in depth). #3353 can be closed after both land.

Requesting human review when convenient.

-- spawn-refactor/pr-maintainer

[la14-1]

Dedup update:#3376 (opened Apr 30) supersedes this PR — it also modifies telemetry.ts with a root-cause fix (runtime env-var guard in sendEvent()) plus adds SPAWN_TELEMETRY=0 in preload.ts for defense-in-depth. Both PRs fix the same leak but #3376 is more comprehensive and will conflict with this one on telemetry.ts.

-- refactor/pr-maintainer

[la14-1]

Closing as superseded by #3376, which provides a more comprehensive fix for the same telemetry singleton leak (runtime env-var guard in sendEvent() + SPAWN_TELEMETRY=0 in test preload).

-- refactor/pr-maintainer

[la14-1]

Superseded by #3376.

[la14-1]

Why: 2 tests fail on every bun test run (hetzner-cov, digitalocean-token) because the telemetry test enables _enabled = true globally and never resets it — every subsequent logWarn/logError triggers fire-and-forget fetch() calls that corrupt other tests' callCount-based mocks.

Root cause

telemetry.test.ts deletes BUN_ENV/NODE_ENV and calls initTelemetry() to enable telemetry for testing. Once _enabled is true, it stays true for the entire process. Any logWarn() or logError() in concurrent test files (hetzner, digitalocean) fires sendEvent() → fetch() through the calling test's mock, consuming callCounts.

Fix (2 lines of real change)

1. Export _testHelpers.enabled getter/setter from telemetry.ts
2. Set telemetryTestHelpers.enabled = false in telemetry.test.tsafterEach

This fixes the leak at the source rather than patching each affected test to filter PostHog URLs.

Relationship to #3341 and #3353

Both existing PRs work around the symptom (patching individual test mocks to ignore telemetry URLs). This PR fixes the root cause — once merged, those workarounds become unnecessary and no future test can be affected by telemetry fetch leaks.

Test plan

• bun test passes with 0 failures (2108 pass, 5376 expect() calls)
• Verified on 2 consecutive runs — no flakiness
• biome check passes on changed files

-- spawn-refactor/test-engineer

[la14-1]

Overlap note: This PR fixes the root cause (telemetry _enabled leaking across test files) of the same 2 flaky tests addressed by #3353 and #3365.

• fix(tests): disable telemetry in afterEach to prevent cross-file fetch leak #3358 (this PR) plugs the leak at the source via afterEach reset
• fix(tests): filter PostHog telemetry from fetch mocks #3353 adds URL filtering in affected tests (simpler defensive fix)
• fix(test): resolve 2 flaky tests caused by parallel fetch mock interference #3365 uses URL-based routing with per-domain counters (more robust defensive fix)

Recommended merge order: this PR first (root cause), then optionally #3365 (defense in depth). #3353 can likely be closed if both land.

All CI checks passing. No merge conflicts.

-- refactor/pr-maintainer

[la14-1]

This PR has been open for 3+ days with all CI checks passing (ShellCheck, Biome Lint, Unit Tests, Mock Tests, macOS Compatibility — all SUCCESS). It is mergeable with no conflicts.

This is the root-cause fix for the flaky test issue (telemetry leak). Recommended merge order: this PR first, then #3365 (defense in depth). #3353 can be closed after both land.

Requesting human review when convenient.

-- spawn-refactor/pr-maintainer

[la14-1]

Dedup update:#3376 (opened Apr 30) supersedes this PR — it also modifies telemetry.ts with a root-cause fix (runtime env-var guard in sendEvent()) plus adds SPAWN_TELEMETRY=0 in preload.ts for defense-in-depth. Both PRs fix the same leak but #3376 is more comprehensive and will conflict with this one on telemetry.ts.

-- refactor/pr-maintainer

[la14-1]

Closing as superseded by #3376, which provides a more comprehensive fix for the same telemetry singleton leak (runtime env-var guard in sendEvent() + SPAWN_TELEMETRY=0 in test preload).

-- refactor/pr-maintainer

[la14-1]

Superseded by #3376.

[la14-1]

Why: 2 tests fail on every bun test run (hetzner-cov, digitalocean-token) because the telemetry test enables _enabled = true globally and never resets it — every subsequent logWarn/logError triggers fire-and-forget fetch() calls that corrupt other tests' callCount-based mocks.

Root cause

telemetry.test.ts deletes BUN_ENV/NODE_ENV and calls initTelemetry() to enable telemetry for testing. Once _enabled is true, it stays true for the entire process. Any logWarn() or logError() in concurrent test files (hetzner, digitalocean) fires sendEvent() → fetch() through the calling test's mock, consuming callCounts.

Fix (2 lines of real change)

1. Export _testHelpers.enabled getter/setter from telemetry.ts
2. Set telemetryTestHelpers.enabled = false in telemetry.test.tsafterEach

This fixes the leak at the source rather than patching each affected test to filter PostHog URLs.

Relationship to #3341 and #3353

Both existing PRs work around the symptom (patching individual test mocks to ignore telemetry URLs). This PR fixes the root cause — once merged, those workarounds become unnecessary and no future test can be affected by telemetry fetch leaks.

Test plan

• bun test passes with 0 failures (2108 pass, 5376 expect() calls)
• Verified on 2 consecutive runs — no flakiness
• biome check passes on changed files

-- spawn-refactor/test-engineer

[la14-1]

Overlap note: This PR fixes the root cause (telemetry _enabled leaking across test files) of the same 2 flaky tests addressed by #3353 and #3365.

• fix(tests): disable telemetry in afterEach to prevent cross-file fetch leak #3358 (this PR) plugs the leak at the source via afterEach reset
• fix(tests): filter PostHog telemetry from fetch mocks #3353 adds URL filtering in affected tests (simpler defensive fix)
• fix(test): resolve 2 flaky tests caused by parallel fetch mock interference #3365 uses URL-based routing with per-domain counters (more robust defensive fix)

Recommended merge order: this PR first (root cause), then optionally #3365 (defense in depth). #3353 can likely be closed if both land.

All CI checks passing. No merge conflicts.

-- refactor/pr-maintainer

[la14-1]

This PR has been open for 3+ days with all CI checks passing (ShellCheck, Biome Lint, Unit Tests, Mock Tests, macOS Compatibility — all SUCCESS). It is mergeable with no conflicts.

This is the root-cause fix for the flaky test issue (telemetry leak). Recommended merge order: this PR first, then #3365 (defense in depth). #3353 can be closed after both land.

Requesting human review when convenient.

-- spawn-refactor/pr-maintainer

[la14-1]

Dedup update:#3376 (opened Apr 30) supersedes this PR — it also modifies telemetry.ts with a root-cause fix (runtime env-var guard in sendEvent()) plus adds SPAWN_TELEMETRY=0 in preload.ts for defense-in-depth. Both PRs fix the same leak but #3376 is more comprehensive and will conflict with this one on telemetry.ts.

-- refactor/pr-maintainer

[la14-1]

Closing as superseded by #3376, which provides a more comprehensive fix for the same telemetry singleton leak (runtime env-var guard in sendEvent() + SPAWN_TELEMETRY=0 in test preload).

-- refactor/pr-maintainer

[la14-1]

Superseded by #3376.