revert(quality): restore fleet CI — #161 made quality.yml unresolvable, every caller startup-failed with ZERO jobs

[rubenvdlinde]

The outage

Every code-quality run created across the fleet after 2026-08-05T13:29:41Z — the timestamp of d68fb727 (#161) — starts up with zero jobs and reports failure. Reproduced on openregister, opencatalogi, openconnector and docudesk, on branches with no changes under .github/.

The two discriminators

This does not look like a broken pipeline from the rollup. It looks like an ordinary red, which is why it is dangerous:

1. The run's name is the raw path — .github/workflows/code-quality.yml — instead of the workflow's name: (Code Quality). GitHub never resolved the workflow, so it never learned its name.
2. Jobs = 0, while conclusion is failure. Anyone reading only the conclusion sees a normal failure and starts debugging their own diff.

Also: gh run rerun refuses it ("workflow file may be broken"), and created_at == updated_at.

Same repo, same callers, minutes apart:

created	run name	jobs
13:22:46Z	Code Quality	28
13:33:34Z	.github/workflows/code-quality.yml	0

Why a revert rather than a forward fix

The cause is above YAML syntax — the file parses cleanly, including under a strict loader with duplicate-key detection — so the fix needs a controlled experiment rather than a guess, and the fleet cannot stay dark while that happens. #161's measured work is good and should return; this only removes it from main until it can be re-landed in a form GitHub resolves.

The change added two workflow_call inputs (43 → 45) and grew the file 207,576 → 224,420 bytes. Both are candidates and neither is yet proven.

Verification before merge

A caller re-run must produce a run whose name is Code Quality with a non-zero job count. A conclusion of failure is not sufficient evidence either way here — the job count is the discriminator.

Refs #155.