A check stopped for CPU reads as the forge's failure, is re-checked once on its own, and never sends its builder a check_failed #11617

Part of #11573 (split from it). Waits for #11613 (cells are stopped only past twice their cores for five minutes) and the worker cap, so such a stop is rare and genuinely the infrastructure's.

Done looks like:

• [x] When a check ends with the watch's CPU stop line (cpuBreach in src/apps/contain.ts), the check records it as an infrastructure failure, re-runs the same head once by itself, and does not send the builder check_failed for that attempt.
• [x] /flow's one sentence names "checks stopped for CPU" when it is the lane's main failure in the last hour.
• [x] A test: a check whose first attempt ends with that stop line is re-queued once and no check_failed is sent; a second such stop is reported normally.
• [ ] After release, the pull pastes the count of production checks in one hour that ended "stopped: used … over its limit": it must be 0. Pulls closed unmerged for this stop since 2026-11-11 21:00Z (#11546, #11547 and any others found) are listed in the pull for the steward to re-queue.

Reviewed pull #11658 at its head: ticked When a check ends with the watch's CPU stop line (cpuBreach in src/apps/contain.ts), the check records it as an infrastructure failure, re-runs the same head once by itself, and does not send the builder check_failed for that attempt.; /flow's one sentence names "checks stopped for CPU" when it is the lane's main failure in the last hour.; A test: a check whose first attempt ends with that stop line is re-queued once and no check_failed is sent; a second such stop is reported normally.. left open: After release, the pull pastes the count of production checks in one hour that ended "stopped: used … over its limit": it must be 0. Pulls closed unmerged for this stop since 2026-11-11 21:00Z (#11546, #11547 and any others found) are listed in the pull for the steward to re-queue. — The count can only be taken after release; the pull lists the pulls to re-queue but the post-release zero is not yet shown..

Left open after pull #11658 landed: 1 unticked item remains (After release, the pull pastes the count of production checks in one hour that ended "stopped: used … over its limit": it must be 0. Pulls closed unmerged for this stop since 2026-11-11 21:00Z (#11546, #11547 and any others found) are listed in the pull for the steward to re-queue.). The planner files each as its own issue and closes this one; a person may close it, tick what is done, or say what is left. Until then the crew does not pick it again.