fix: remove global-setup REST warm-up that hung CI for 30 minutes

[josephfusco]

The last main run (from #28's merge) ran "Run Playwright tests" for 26 minutes with zero test output ever produced, hitting the job's 30-minute cap and getting force-killed. Traced this to the RequestUtils REST warm-up I added to global-setup.ts in a previous PR: it's the first thing that runs, and something about it hangs indefinitely in real CI (never reproduced this locally, which is exactly the risk of trusting an untested-in-CI addition).

The two actual fixes from this session (Gutenberg auto-enabling the collaboration experiment, and the corrected filter-detection notice) don't depend on this warm-up at all. Verified locally against a genuinely cold-started wp-env: 11/11 passing in ~22s with it removed entirely.

Use of AI Tools

Assisting with diagnosing the hang from CI logs and implementing the revert.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 63.90%. Comparing base (8d09690) to head (07459c0).

Additional details and impacted files

@@            Coverage Diff            @@
##               main      #30   +/-   ##
=========================================
  Coverage     63.90%   63.90%           
  Complexity       35       35           
=========================================
  Files             7        7           
  Lines           266      266           
=========================================
  Hits            170      170           
  Misses           96       96           

Flag	Coverage Δ
multisite	63.90% <ø> (ø)
phpunit	64.63% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ 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.

[josephfusco]

Playwright Tests had three separate causes, fixed in order:

1. Slow user provisioning. ensureCollaborator() created test users over REST (admin.createUser()), which pays a full login + REST-discovery bootstrap per call. Switched to WP-CLI (wp user get/create/update), ~1s instead of 30s+.
2. Wrong WordPress instance. getAPIRootURL() in @wordpress/e2e-test-utils-playwright resolves via its own internal WP_BASE_URL default (:8889, wp-env's "tests" env) whenever process.env.WP_BASE_URL is unset, ignoring our configured baseURL. .env is gitignored and never exists in CI, so every REST call was silently hitting the wrong WordPress instance and getting 403s. Confirmed via the failed run's trace.zip network log. Fixed by committing .env.example and copying it into place in CI.
3. Slow/flaky browser install. Install Playwright browsers ran apt-get on every run regardless of cache hit, and the cache key had no restore-keys fallback. A lockfile-hash cache miss plus a stalled apt mirror hung the job 20+ minutes. Added the fallback and skip apt-get on cache hit.

Verified locally against a cold wp-env restart and the CI-exact .env-missing condition before each push.