Fix e2e cleanup

[isaacroldan]

Why

The e2e tests create temporary Shopify apps in the Dev Dashboard. Each app name includes the GitHub Actions run token, for example E2E-scaf-rcyjrwc3a1-mqu8cgqh, so cleanup can find everything created by one run.

A previous run left apps behind even though the cleanup step ran. The per-test delete flow failed on a Dev Dashboard error page, then the fallback cleanup opened the apps list and saw 0 app cards. Because the script treated that as a real empty result, it exited successfully instead of retrying or failing.

What

This makes the fallback cleanup more reliable:

• It opens the Dev Dashboard apps page with search_term=<run token>, so the page only returns apps from the current e2e run instead of scanning every app in the org.
• It still checks the returned cards locally before deleting anything.
• It accepts the dashboard's real empty-search state (No apps matched your search.) as a valid zero-result cleanup.
• It rejects broken dashboard states, such as error pages or pages with no app cards and no empty-search message, and retries before failing.

Testing

• pnpm --filter e2e exec tsx scripts/cleanup-apps.ts --list --pattern rcyjrwc3a1
• pnpm --filter e2e exec tsx scripts/cleanup-apps.ts --list --pattern rcyjrwc3a1ss
• pnpm --filter e2e type-check
• git diff --check

[isaacroldan]

• Fix e2e cleanup #7936 👈 (View in Graphite)
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.