CI fleet: runs are starving (queued 30+ min)

[github-actions]

Runs have been queued past the 30-minute threshold — jobs are not being picked up.

• Kani on dependabot/cargo/fuzz/wit-bindgen-0.62.0 — queued 51 min (run)

Wedged runners (offline while holding a job): none detected / inventory unavailable

This is usually the self-hosted fleet being down, NOT a code fault: jobs queue forever against a label with no live runner, which looks identical to a slow build.

Detected by: https://github.com/pulseengine/relay/actions/runs/35263935964

[avrabe]

False positive, measured — the check reads RUN status, not jobs (#449)

The alarm says Kani on dependabot/cargo/fuzz/wit-bindgen-0.62.0 was "queued 51 min". The run's own jobs say otherwise:

• run 35258925212 created 18:27:24Z
• its first job started 18:32:22Z — 5 minutes later
• 47 jobs in total, every one of them started

The run stayed in the queued run status while its matrix legs were staggering through the fleet, which is the exact defect #449 describes: starvation is measured from run status, so a long matrix looks identical to work nobody picks up. The fleet was not starving — it was busy with the Kani soak re-runs for #451 and three PRs' checks.

This is the second false positive of this shape (#447 was the first). Closing; the fix is tracked in #449.