Restore what my stale-branch squash silently reverted in witness_floor_workflow

[gunbai-bot]

#8702 (mine) silently reverted another session's work. Not a conflict, not a test failure — a clean squash-merge of a stale branch.

It deleted a 14-line annotation block and reverted a CI step name that #8657 had authored. I did not write those deletions. My branch predated #8657; squash-merge takes the branch's version of every touched file wholesale, so the result presented as if I had authored the removal.

What was lost

• The annotation explaining why the behavioral receipt is not a step here — that CI: four job steps become one invocation, four in-process phases #8647 collapsed the step ladder into one invocation, that re-adding steps would rebuild the ladder that PR removed, and that the per-PR phase now decides from what it can observe rather than from a trigger name. That last paragraph records a behavioural difference, not a relocation, and it is what a future reader needs and cannot re-derive.
• The step name Required CI: parse, regen, regen determinism, behavioral receipt, witness floor.

Why nothing caught it

1. Squash-merge of a stale branch presents a revert as an authored deletion.
2. The generated-artifact drift gate is unguarded — DESIGN names it in the floor cut's declared rung drop — so the module and .github/workflows/witnesses.yml disagreed on main with nothing to notice.
3. The merge was clean, five reviews approved, and CI passed.

Found only by chasing a 20-byte mismatch while byte-comparing an emitted artifact in an unrelated branch — precisely the check CI is currently missing.

No regeneration required

Under the fleet stop-the-line rule this matters: the committed artifact still carries the correct text, so restoring the module makes the two agree again.

check	result
witnesses.yml emission	IDENTICAL 1457 == 1457 (was 1437 vs 1457 on main)
the_live_witness_floor_job_closes_its_capabilities	true — #8702's binding survives
in-body annotations	0

The general hazard

A long-lived branch plus squash-merge is a silent-revert machine. Every hour a branch sits unmerged, its copy of each touched file becomes a stale snapshot that overwrites whatever landed meanwhile — and no conflict, no review, and no green CI will say so while the drift gate is down.

🤖 Generated with Claude Code

[gunbai-bot]

CI failure at dfeb66e is the same pre-existing main breakage as on #8734, not a defect in this PR:

required-ci: FAILED PHASE floor refused: subject=1941a771dcc5ce65 modules_resolved=3760 modules_excluded=4
dag/gunbc/fleet_desired_admission.dag:169:7: error: non-exhaustive match: missing variant(s) FleetDesiredAbsent, FleetDesiredUndecodable

dag/gunbc/fleet_desired_admission.dag is untouched by this PR. #8701 split FleetDesiredRevision from two variants to four in gunbc.fleet_main_revision without answering the two new arms at the fleet_desired_admission_decide consumer; main's witness floor has been red on every run since that merge (73c62cad34 onward) and every PR that merges main inherits it.

#8738 is the fix and is open and MERGEABLE. Merge order: #8738 → this PR → #8734.