Brief: P7 edge-door plan, review round 2

Round 1's brief is CODEX-REVIEW-BRIEF-P7-EDGE-DOOR.md and its findings are CODEX-FINDINGS-P7-EDGE-DOOR.md, both beside the plan. The plan, docs/superpowers/plans/2026-09-28-instance-moves-p7-edge-door.md, is now revision 2. It opens with a table giving each round-1 finding's disposition.

1. Disposition table first. For each of the ten round-1 findings, say in one line whether revision 2's fix actually closes it (closed / partly / not closed), and why where it isn't closed.
2. Then only new findings, ranked P1/P2/P3, each with the failure scenario and a fix. Look especially at what revision 2 introduced: - the console database as authority and the unconditional publish; - the revision high-water state file, and its behaviour on a fresh box, after data loss, and when the console's revision counter restarts; - the door-hop MAC and the door capability flag in edges files; - enforcement moved into ServeHTTP, including the ordering against the tunnel and steering branches and the re-read of the door mark after acceptHop; - openDoor lifecycle placement.
3. The rulings in the design record's §0 are fixed. Don't re-argue them, or round-1 dispositions marked accepted, unless the fix itself is wrong.
4. Check every path:line you cite against this checkout.

End with one line saying whether the plan is ready to build. Keep it under about 1,500 words.