[Phase 2] SC2 game-day: prove a broken next cannot block its own fix PR

[don-petry]

Parent: #495 · Phase 2 · Work type: Spike · Targets SC2 (must-have).

Game-day: deliberately ship a broken next version of pr-review/dev-lead and prove the fix PR still gets reviewed, approved, and merged — because production duty runs the pinned stable. Codify as a regression guard.

Done when: demonstrated + a regression test asserts the circular dependency stays broken.

[github-actions]

Thanks for opening this — proving that production's pinned stable duty breaks the circular dependency is a great resilience property to lock in. To help scope the regression guard: do you envision it as a dedicated CI workflow (e.g. a workflow that spins up a broken next and asserts the fix PR still reaches merge), a shellcheck/bats-style test in the scripts/ directory, or something else?