source-control:babysit-prs: configured-login clean-approval material opt-in is unreachable (contradicts PR #567's own doc contract)

[kyle-sexton]

Lane: babysit. Triggering PR: #567 (fix/499-babysit-snapshot-approve-nits-classification).

Worker's live re-review of the pushed fix (c86fc59) found a third codex P2: "surface configured-login clean approvals as material" is contradicted by the current code — the opt-in path for treating a clean approval from a configured reviewer login as material spans two separate routing paths, and as shipped the configured-login set is empty/zero users by default, making the documented contract unreachable in practice. Fixing requires either a real code fix across both routing paths, or tightening the doc to match actual (currently unreachable) behavior — an author-owned call, not mechanical.

Category: bug/behavioral (doc-vs-code contradiction). Needs a human decision: code-fix vs doc-tighten.

[kyle-sexton]

This was generated by AI during triage.

Umbrella-fold → #534 (babysit merge-readiness classifier). "Configured-login clean-approval opt-in unreachable" is approval-state classification — same pattern as #594, same umbrella. Not routed standalone; blocked by #534 recorded natively.

[kyle-sexton]

Control-tower close (2026-07-20, merge-digest evidence at PR #567 head 348d3fa). Premise removed: (1) the two-paths concern is consolidated — the only routing reference to approval_downgrade_logins is the single elif approval_downgrade(text): branch (babysit_feedback.py:398-417) with the material-vs-ignored decision colocated (:412); (2) the doc contradiction is resolved by narrowing — a PLAIN clean approval falls through to the terminal else: ignored (:425) which never consults the config, and the shipped docs now say exactly that; the material opt-in fires only for the approval-carrying-blocking-prose case. No residual code gap. Reopen if a live counter-example surfaces.