feat(ci): integrate Datadog flaky test detection
- Query Datadog CI Visibility API for historical test data
- Identify tests with flaky behavior on canary branch (30 day window)
- Split failures into 'Needs Investigation' vs 'Known Flaky'
- Show flake rate and recent failure count for known flaky tests
- Known flaky tests collapsed and deprioritized in comment
- Uses existing DD_APPLICATION_KEY secret for API access