Follow-ups from #971 review: make the LambdaGuard control catch a non-blocking harness #987

#971 (issue #962) merged at 83459e4. Its independent review left three non-blocking notes:

1. In background_collection_test.go:161, deleting <-release still passed with -count=100, so the positive control wouldn't reliably catch a harness regression.
2. The Lambda subtest's time.After(200ms) adds about 0.2s on the pass path.
3. It's not stated which test pins which isLambda guard (scheduler.go lines 1183, 1452 and 1502).

The follow-up commit was written after #971 merged, so a separate PR carries it.

🤖 Generated with Claude Code