[jacksteamdev]

Summary

Preserve analytics capture across reload in the browser test that blocked release PR #33.

- remove all request interceptors and reinstall measurement capture
+ remove only the temporary birth-resolver failure handler
+ verify eligibility events belong to two distinct visits

Also format one existing YAML string so the current main branch passes formatting checks.

Evidence

• Before: PR #33 preview run failed with one captured eligibility event instead of two. Its trace confirms the second event reached the server but bypassed test interception.
• After: Type checks, lint, formatting, content checks, six script tests, production build, and staging/alpha deployment dry runs pass. Regression passed 10/10 local repetitions. GitHub preview E2E passed the full 43-test suite. A separate run against the old preview passed 42/43 with a navigation timeout; that test passed locally on rerun. GitHub CI checks passed. The independent Claude review check failed with "No new marked Claude review for triggering SHA".

Merge Danger

Door: two-way

Blast Radius: tests

Changes browser-test routing and YAML formatting; no production behavior changes.

[github-actions]

Review incomplete (action: success; verification: failure). Run log

Scope: full (no-previous-review); commit: eed4f0d2a7f1a67dd2c229837214bb6d6fc0ea44.