chore(priority-bump): remove the dead inputs, describe what it actually does (backend#1597)

[LukasWodka]

Item 2 of tracebloc/backend#1597 (source finding: backend#1408 P3 — customer-priority-bump.yml:9-20 has three dead inputs and the step at :37 hard-codes the priority label while the org docs still say Priority = P1).

What was dead

Three of the four workflow_call inputs were declared with defaults and never referenced anywhere in the file:

input	default	references in the workflow
project-number	2	0 — removed
org	tracebloc	0 — removed
target-priority	"P1"	0 — the step hard-codes --add-label priority; removed
trigger-label	from:customer	1 (job if:) — kept, unchanged

They are leftovers from the pre-D5 design that wrote the board's Priority field. Under D5 the workflow adds a binary priorityissue label instead, and the Priority single-select no longer exists on the board (verified against project 2's live fields: Status / Work Type / Deploy environment / Area only). Removal over wiring: making the label name configurable would be redesign, and the ticket asks for cleanup.

The header comment now states exactly what the workflow does — trigger label in (default from:customer), binary priority label out, no board field written — so the file no longer needs its inputs block to be read sceptically. Deliberately not described as "sets Priority=P1 on the kanban card": that is the stale side of the #1408 doc-alignment finding. The org CLAUDE.md workflow table still says Priority = P1 and needs a docs-side fix outside this repo; repo-inventory.yml's customer_priority_bump_caller_missing note can drop its dead-inputs clause once this is promoted.

Caller-payload measurement (why outright removal is safe)

Removing an input a caller still passes would fail the call, so every caller was measured first: all 16 repos with a customer-priority-bump.yml caller (averaging-service, backend, claude-skills, cli, client, client-runtime, data-ingestors, design-system, docs, e2e-test-agent, frontend-app, model-zoo, start-training, tracebloc-engine, tracebloc-py-package, tracebloc-website), each read on both develop and its prod branch (main/master). All 32 copies are byte-identical (sha1 prefix a04b231c): uses: …@main + secrets: inherit, zero with: blocks. No caller passes any input, so the deprecated-no-op fallback is unnecessary and nothing can break — including at @main promotion time, since callers pin @main and prod-branch callers are the same bytes.

Behavior unchanged

• Same trigger condition: github.event.label.name == inputs.trigger-label, default from:customer.
• Same action: gh issue edit "$ISSUE_NUMBER" --add-label priority under PROJECTS_KANBAN_TOKEN.
• The run: step and job are byte-for-byte untouched; only the header comment and the three unused input declarations changed.
• YAML parses; actionlint passes.

Part of tracebloc/backend#1597

🤖 Generated with Claude Code

---

Note

Low Risk
Documentation and dead-input removal only; trigger logic and the priority label step are unchanged and callers pass no removed inputs.

Overview
Cleans up the reusable customer priority bump workflow by dropping three unused workflow_call inputs (project-number, org, target-priority) left over from when the flow wrote a kanban Priority field. Only trigger-label remains (default from:customer).

The top-of-file comment now documents the real behavior under D5: when the trigger label is added, the job adds the binary priority issue label and does not set any project board field. The gh issue edit --add-label priority step and job conditions are unchanged; callers that use secrets: inherit with no with: block are unaffected.

^{Reviewed by Cursor Bugbot for commit de786b7. Bugbot is set up for automated code reviews on this repo. Configure here.}