[codeforester]

Summary

Make the timeout-clock no-retry regression test independent of child-process scheduling. The fake timeout helper now counts its actual executions after the --version probe, and the test asserts it ran once while preserving the infrastructure-failure diagnostics. Production timeout and retry behavior do not change.

Issue

Fixes #583

Validation

• Four focused timeout/supervision BATS tests passed.
• The repaired regression passed ten consecutive runs.
• ./tests/validate.sh passed: 712 BATS tests, 26 Project Intake tests, and all repository artifact, API, compatibility, release, integration, quality, and CI-policy checks.
• git diff --check passed.

Demo Impact

None.

Docs Impact

None.

API Impact

None. Production code and public behavior are unchanged.

CI Impact

The timeout-clock failure regression now measures clock invocations directly and does not depend on the supervised child running before the process group is stopped.

Security Notes

None.

Notes

This carries forward the independent test repair from closed PR #582. The curated .ai-context/ files and related contribution guidance are excluded.

Checklist

• Branch follows <category>/<issue>-<YYYYMMDD>-<slug>.
• Pull request is scoped to one issue and links it with Fixes #583.
• Validation is recorded above.
• Applicable impact sections are complete.
• No user-facing or release-worthy behavior changed; no CHANGELOG update is needed.