[playwright] Add support for handling option preferences and reactivate skipped test

[planger]

What it does

Due to the Monaco uplift, we had to temporarily skip a test, because the Monaco change affected the default preference regarding auto-save.

To re-enable the skipped test, we now explicitly set the auto save preference to off before the test. This way the preference value is ensured to be consistent. In order to do that conveniently, this change adds general support for preferences that are set by HTML select/options. In the course of introducing support for select/option preferences, we harmonized the code for handling different types of preferences in the TheiaPreferenceView page object.

See also #10736
Fixes #10891

How to test

This change adds additional tests for the new support for option preferences and reactivates the test that had to be skipped in #10736.

Review checklist

• As an author, I have thoroughly tested my changes and carefully followed the review guidelines

Reminder for reviewers

• As a reviewer, I agree to behave in accordance with the review guidelines

[colin-grant-work]

A few of the tests failed for me (MacOS) when I ran this PR alone, but they passed when I rebased this work on #10925. Let's merge that one, rebase this one, and confirm that everything is working once that's done.

[planger]

Thanks for testing, @colin-grant-work! #10925 is now merged and I rebased this PR. Build is still running, but the Playwright Tests and one CI/CD Build is green already.

[planger]

Hm, one build reported two failing unit tests:

root INFO   821 passing (3m)
root INFO   2 failing
root INFO   1) Launch Preferences
       Valid Compound Launch Configuration
         ".theia"
           delete launch:
     AssertionError: expected null to deeply equal undefined
      at Context.<anonymous> (/home/runner/work/theia/theia/examples/api-tests/src/launch-preferences.spec.js:658:24)

root INFO   2) Launch Preferences
       Valid Compound Launch Configuration
         ".vscode"
           delete launch:
     AssertionError: expected null to deeply equal undefined
      at Context.<anonymous> (/home/runner/work/theia/theia/examples/api-tests/src/launch-preferences.spec.js:658:24)

Seems unrelated on a first sight. Is this a flaky test as it passes on node 12 and node 16 if I'm not mistaken?

[colin-grant-work]

@planger, that's on the list of known flaky tests, so intermittent failures in the Launch Preferences suite can mostly be ignored. The undefined !== null complaint is relatively recent, but I've also made note of that specifically, so it's definitely nothing to worry about. I'll take a look at some of the flaky tests after the release.

[planger]

I had to adjust the preference editor page object as we don't use select elements anymore for options, but render them with a div and a popup div on latest master. Now the tests are green again and the previously skipped test is re-enabled.

Can you please take a look if those changes are fine with you? Thanks!

[vince-fugnitto]

I confirm that the tests pass during CI and locally 👍

[vince-fugnitto]

LGTM 👍 The playwright tests pass successfully.

[JonasHelming]

@planger Could you squash please, I am not sure how the final commit message should look like

[JonasHelming]

@colin-grant-work : Should I wait merging this?

[colin-grant-work]

@JonasHelming, no; it's good to go. I was enqueuing my reviews this morning, and I hadn't noticed that Vince had already approved. I'm fine with what I've seen with the changes, and his is good enough for me.