test: add front-end tests for react code

[afuetterer]

Description

This PR proposes the following changes:

• add pytest-playwright as test dependency
• add a custom e2e pytest marker, disabled by default
• add django debug mode selection per env (e2e tests are run with DEBUG=True)
• add e2e tests for management navigation
• remove dedicated "webpack-build" CI job and build frontend bundle in test job
• refactor fixtures fixture to be reusable by other tests and fixtures
• test cases for the management UI

How to run e2e tests

Run the e2e tests with the custom marker:

pytest -m e2e

Run the e2e tests with an actual visual browser:

pytest -m e2e --headed

Slow down the test execution:

# --slowmo: miliseconds between actions
pytest -m e2e --headed --slowmo 500

Tests suggested by @MyPyDavid in #715 and related tasks

• test the "Management Interface"
• test navigation in #sidebar
• count number of items per element type (in #list-group), as per number of items in the db (i.e. fixtures)
• decide if e2e tests should run once or per Python/DB combination: yes, run only once
• decide if e2e tests should run with a single browser: chromium
• rebase branch
• refactor tests to minimize code duplication -> parametrize
• test nested display for functionality (eg. /catalogs/1/nested/)
• cache browser installation in CI
• rebase branch
• Python 3.12 playwright (waiting for 1.39 on PyPI, already available on GitHub)
• run tests in parallel in CI with xdist: django.db.utils.OperationalError: database table is locked
• test new item creation, fill data, save in db
• test edit buttons and edit form for each element type
• squash commits
• add section to testing docs: docs: add section about e2e tests rdmo-docs#35
• remove nested_elements
• fix: update verbose_name_plural in questionset #817 is merged
• rebase branch
• refactor test_management_navigation() to use a for loop through model_helpers
• remove properties from ModelHelper
• squash commits

Related issue: #715

Types of Changes

• Refactoring (no functional changes, no api changes)
• Build related changes

Checklist

• I have read the contributor guide.
• My code follows the code style of this project.
• My change requires a change to the documentation.
• I have updated the documentation accordingly.
• I have added tests to cover my changes.
• All new and existing tests passed.

[MyPyDavid]

I've tested it out with pytest -m e2e --headed, that looks pretty awesome and really fast ;)
Think that screenshots would be overkill for the moment, could use it maybe when there are multiple developers that work on features for the front-end.

[afuetterer]

Apropos fast, you can add slomo.

See: https://playwright.dev/python/docs/test-runners#cli-arguments

[MyPyDavid]

ah yes, thanks for the snippet!
sure, I can look it at, maybe I'll put a sleep in between ;)
until later then!

[afuetterer]

Playwright usually waits, you can also increase its timeout settings. I guess no need for sleeping.

[MyPyDavid]

the test can also be finished via the front-end only:

    page.get_by_title("Edit catalog").first.click()
    comment_locator = page.get_by_label("Comment")
    expect(comment_locator).to_have_text(comment)
    # Catalog
    catalog = Catalog.objects.first()
    # now it passes
    assert catalog.comment == comment

but now also the db assert works..
think that the expect executes the save to the db and before that it is lazily evualated?

[afuetterer]

Awesome, you already found a solution? Nice.
Maybe you also know what kind of stuff people usually edit there. I just wanted to edit anything and hit save. But there is a thousand options to edit.

[afuetterer]

If you want to play around with test generation, please try:
python manage.py runserver in one terminal and
playwright codegen localhost:8000 in another.

[afuetterer]

Done with this for now. Tests are passing.

[MyPyDavid]

that is awesome! thanks so much XD

[afuetterer]

that is awesome! thanks so much XD

You are welcome.

[jochenklar]

Awesome, thanks this looks very good!

[jochenklar]

I don't mind setting this always to True, I don't think this makes the tests much slower.