Our functional test suite currently uses Selenium with gunicorn, Chrome, Firefox, geckodriver and chromedriver. This is somewhat troublesome to maintain:

• Dependencies are pinned to outdated versions (Selenium and urrlib) due to breaking changes that are left unresolved. This also affects integration tests.
• Test infrastructure is somewhat fragile. Gunicorn polls the subprocess until it responds, and tests use a lot of sleep() commands to wait for the frontend to update its state.
• The Docker image installs a lot of dependencies in order to run the headless browser for testing, and requires drivers to interact with the browsers.

Playwright offers a better alternative:

• Auto-waiting allows Playwright to wait for elements to be actionable, and is handled by the library instead of manual sleep calls.
• The test server can be simplified by using the built-in LiveServerThread in Django
• The docker image is lighter because we can use the browser binary from Playwright, without extra browser repositories or drivers.
• Fewer dependencies are needed to run functional tests, and we can remove selenium, gunicorn and whitenoise.

The test suite is small (7 tests across 6 files), so this change is low risk and should make the tests more reliable.