fix(push-hook): retry MR commits fetch to handle GitLab async race condition

[sylvaincombes]

Problem

When Homer receives a push hook, it filters the pushed commits by cross-referencing the GitLab MR commits API — this removes rebase noise by keeping only commits that actually belong to the MR.

The issue: GitLab updates MR content asynchronously. Right after a push, the just-pushed commit is sometimes absent from the MR commits API response at the time Homer fetches it. The intersection is empty, the handler silently returns 204, and no Slack notification is sent. No error is logged, making it invisible.

Fix

Introduce waitForMergeRequestCommits, a utility that polls the MR commits endpoint until at least one pushed commit appears, then returns. It follows the same pattern as the existing waitForReleasePipeline.

• Retries every 1s up to a 5s timeout
• Transient fetch errors (e.g. GitLab 5xx) are caught and retried rather than aborting
• Falls back to the last known result on timeout, preserving the existing silent-204 behaviour as a last resort
• Logs a warning on timeout so the failure becomes visible

Trade-offs

• The webhook response is now delayed by up to ~5s on the stuck path. GitLab's default webhook timeout is 10s so this should be safe, but worth monitoring.
• A cleaner long-term design would be to ack the webhook immediately (200) and process asynchronously, which would remove the timeout concern entirely and allow a more generous retry window. Left as a future improvement.

Changes

• src/review/commands/share/utils/waitForMergeRequestCommits.ts — new polling utility
• src/review/commands/share/hookHandlers/pushHookHandler.ts — use the new utility
• __tests__/review/waitForMergeRequestCommits.test.ts — unit tests covering: immediate hit, retry-until-found, error retry, timeout
• __tests__/review/pushHook.test.ts — one-line update: the "no related review" test now correctly gets its 204 from no review in DB (the retry logic is owned by the unit tests above)

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 89.36%. Comparing base (9315d06) to head (0b784dc).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #174      +/-   ##
==========================================
+ Coverage   89.29%   89.36%   +0.07%     
==========================================
  Files         104      105       +1     
  Lines        2269     2295      +26     
  Branches      440      444       +4     
==========================================
+ Hits         2026     2051      +25     
- Misses        230      231       +1     
  Partials       13       13              

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.