fixing-flaky-e2e-tests

Run the fetch_flaky_tests.py script, which downloads playwright_test_stats artifacts from recent successful CI runs and aggregates tests that required reruns. Use --days, --top, and --min-reruns options to control the lookback window and result count.

Compare actual versus expected images in the test-results/snapshot-updates directory to identify timing, animation, or rendering causes. Add explicit waits such as expect(element).to_be_visible() or wait_for_images_loaded() before taking the snapshot.

Firefox has slower console logging, request retries, and subpixel rendering differences, while WebKit has layout timing variations. Fixes include longer timeouts, waiting for image decode, or adding a markdown element to stabilize Firefox subpixel positioning.

Skipping with skip_browser is a last resort, acceptable only after exhausting timing fixes and confirming the flakiness stems from a documented browser limitation rather than a test bug. A clear reason must be included in the marker.

If the test passes all 25 local reproduction runs, do not attempt a fix. The flakiness likely stems from CI environment factors that cannot be addressed locally, so skip it and move to the next test.