tools/check_seal_coverage.py is one comment line on master: a bee merge her verdict refused deleted the seal gate #5453

tools/check_seal_coverage.py on master is one line: # No changes made to this file; only the baseline was updated. It was 688 lines until #5183 (2026-09-29, branch queen-5158) replaced it. An empty Python file exits 0, so the coverage job and its --self-check negative control have reported success since then while checking nothing.

• The Queen's verdict on 131 seals are stale: the spec changed after sealing #5158 is escalate; she did not accept it. It merged under the publisher of that day, which armed every bee PR; feat(queen): a bee branch merges only on the Queen's accept at its head #5422 closes that road.
• 131 seals are stale: the spec changed after sealing #5158 itself said re-sealing the stale seals is "not mine": it is the owner's call. The merged change answered it by deleting the gate.
• Measured on 2026-10-01 with the 688-line checker restored, at master, with t27c built from master's bootstrap/: its self-check passes, and the real run reports 649 seals that do not hold (540 stale, 109 gen-drift). That was 131 when the gate was removed.

Wanted: restore the checker byte-for-byte from bcb32d723^. Re-sealing stays a separate owner decision, so the restored gate will honestly report red until that decision is made.

(No boundary section on purpose: this is being done in a PR already, not by a bee.)