docs(qa): correct two dogfood headers that still state ADR-0094's retired 2026-07-14 overlay direction

[os-zhuang]

Part of #7351 — this is the packages/qa half. The card's third site (the
ADR-0094 D2 parenthetical, under docs/adr/**) ships as a separate PR, because a
docs/adr/** path hit routes a PR to maintainer-only landing and this half needs no
such wait. Neither PR blocks the other.

What was wrong

ADR-0094's 2026-07-14 direction — an env-scope overlay is the platform's standard
ADR-0005 customization of a packaged permission set, and deleting the overlay resets
the row to the shipped declaration — was retired on 2026-08-09 by ADR-0094 D5-R
(#6858 / PR #6962), after #6483 / PR #6608 rolled permission back to
allowOrgOverride: false.

Two dogfood file headers still described that retired direction as current:

• packages/qa/dogfood/test/two-doors-permission.dogfood.test.ts — the 块2 header
  stated both retired halves verbatim ("TRANSLATED into an env-scope metadata OVERLAY
  (the standard ADR-0005 customization)" and "'delete' resets to the shipped
  declaration").
• packages/qa/dogfood/test/showcase-permission-projection.dogfood.test.ts — header
  item 3's looser "deleting an artifact-backed set RESETS it to the declared body".

In both files the test bodies are already correct and carry their #6483
annotations; only the headers had drifted. This is the same silent-by-construction
class as #7082, one package over.

What changed

Header prose only. Each header now states the refusal that its own body asserts, marks
the 2026-07-14 direction as retired, and points at D5-R. The showcase header additionally
records that with no overlay left to lift the delete is a no-op success rather than a
revert-to-declared step — matching the annotation its delete case already carries.

Not touched, on purpose: no assertion, no behaviour, no gate. The card is explicit
that the test bodies are right and the header is what is wrong, so nothing was "fixed"
to match a stale header. No currency gate was built (that idea is #7082's disposition B,
which was not dispatched).

Verification

Premise re-verified before editing: the card was measured on origin/main @ f3f855ac,
so all three quoted passages were re-located by text and confirmed still verbatim on
ecb39ea22.

• The diff is mechanically comment-only — every changed line is a // comment
  (git diff -U0 filtered for non-comment changes returns empty).
• pnpm --filter '@objectstack/dogfood^...' build (62-package closure), then
  pnpm --filter @objectstack/dogfood typecheck — clean.
• pnpm --filter @objectstack/dogfood test — 104 test files passed, 1 skipped; 719
  tests passed, 2 expected-fail, 3 skipped, 0 failed.
• Gates re-derived against the final diff with node scripts/pm/dispatch-gates.mjs:
  check:test-source-alias, check:type-source-resolution (path-derived),
  check:query-options-erasure, check:type-check-coverage (convention-triggered by
  editing test files), plus check:nul-bytes. All five green locally.

No changeset: comment-only change in test files, nothing user-visible is released — the
PR carries skip-changeset.

---

Generated by Claude Code

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
objectstack	Ignored		Aug 13, 2026 3:34am

[github-actions]

📓 Docs Drift Check

No hand-written docs reference the 0 changed package(s). ✅