Playwright automated functional test suite — CI integration

Summary

Wire the new Playwright pytest suite into CI so it runs on every PR.

Test file: tests/ui_functional/test_ui_functional.py
Conftest override: tests/ui_functional/conftest.py

Current state (2026-05-31)

• 62 tests, all passing
• Covers all 5 protected pages × 3 viewports
• Tests all interactive buttons and form actions without Plaid credentials
• Screenshots saved to tests/ui_functional/screenshots/

Tasks

• Add playwright install chromium step to CI workflow (GitHub Actions)
• Add pip install playwright to CI requirements
• Wire into the existing CI job: pytest tests/ui_functional/ -v
• Upload tests/ui_functional/screenshots/ as CI artifact on failure
• Add a @pytest.mark.playwright marker and a -m playwright runbook to TESTING.md
• Extend test suite: add tests for inline rename (ledger, account, line item)
• Extend test suite: add tests for add-item-input (Enter key flow)
• Extend test suite: add tests for account description save (PATCH)

Acceptance criteria

• CI runs Playwright tests on every PR against main
• Failed tests produce screenshot artifacts for debugging
• Test runtime < 90s

Part of epic #299.

Start with tests/ui_functional/test_ui_functional.py and tests/ui_functional/conftest.py, then inspect the existing GitHub Actions workflow and CI requirements. Check TESTING.md for the runbook structure. Done means the Playwright suite runs on every PR, failed runs upload tests/ui_functional/screenshots/, the marker and runbook are documented, and the listed coverage extensions are included within 90 seconds.