Observed during the follow-ups re-gate (2026-08-04), on both pre- and post-#327 trees — NOT caused by that branch: under full-suite parallel load, 1-2 nested-routing tests intermittently fail on chromium, and WHICH tests fail varies between runs (seen: 2, 6, 8, 9). Isolated runs (--repeat-each=3, single project) are consistently green, and full-matrix runs are usually green (two consecutive full runs: one with a single flaky failure, one 521/0). Likely timing sensitivity in route-swap/active-class assertions under worker contention. Needs the same deflake treatment as #277/#278: identify the non-retrying assertion or race in nested-routing.spec.ts tests 2/6/8/9 and replace with deterministic waits.
Observed during the follow-ups re-gate (2026-08-04), on both pre- and post-#327 trees — NOT caused by that branch: under full-suite parallel load, 1-2 nested-routing tests intermittently fail on chromium, and WHICH tests fail varies between runs (seen: 2, 6, 8, 9). Isolated runs (--repeat-each=3, single project) are consistently green, and full-matrix runs are usually green (two consecutive full runs: one with a single flaky failure, one 521/0). Likely timing sensitivity in route-swap/active-class assertions under worker contention. Needs the same deflake treatment as #277/#278: identify the non-retrying assertion or race in nested-routing.spec.ts tests 2/6/8/9 and replace with deterministic waits.