Use CondVar instead of UnitTest.wait in tests

[patrickdupuis]

Purpose and Motivation

Fixes #5366

This PR removes all but one use of UnitTest.wait inside our test suite (the exception being testing that method itself). CondVar is used as a preferred alternative. I've updated the documentation, and added comments and notes pointing users towards CondVar.

I've also re-written the implementation of UnitTest.wait. This method no longer makes use of Condition as this Class is likely going to be deprecated in the future (or at least no longer be used inside the SC class library).

Types of changes

• Documentation
• Bug fix
• Breaking change (maybe?)

To-do list

• Code is tested
• All tests are passing
• Updated documentation
• This PR is ready for review

[dyfer]

Thanks @patrickdupuis !

Could you remove the "skip" entries for the fixed tests from the gha_test_run_proto.json?
E.g. remove the following:

supercollider/testsuite/scripts/gha_test_run_proto.json

Lines 19 to 24 in 6c49680

	{
	"suite":"TestCoreUGens",
	"test":"*",
	"skip":true,
	"skipReason":"UnitTest has bugs (FIXME)"
	},

as well as other fixed tests. Once you do, these tests will run in CI in this PR.

[patrickdupuis]

@dyfer I've updated the list of skipped tests in the CI. Let's see what happens in GA :)

[dyfer]

@dyfer I've updated the list of skipped tests in the CI. Let's see what happens in GA :)

I don't think we can run TestDocument in the current configuration since we are not running the IDE, so these should stay skipped. The only other test that's failing is TestUnitTest:test_setUpClass_already_happened

[patrickdupuis]

I've added those tests back to the skipped list.

[patrickdupuis]

Should we re-run the CI a few times just to make sure everything is passing?

[patrickdupuis]

I cleaned up the commit history. Let's run the CI one last time before merging.

[patrickdupuis]

@dyfer the test that's failing for Linux hasn't happened before. I'd rather just ignore it and merge the PR. What do you think?

[dyfer]

@dyfer the test that's failing for Linux hasn't happened before. I'd rather just ignore it and merge the PR. What do you think?

I'd be fine ignoring it, but I have seen TestSoundFile_Server:test_cue_playNow failing before, albeit rarely. Up to you whether you want to add skipping it, or merging this as is.

[patrickdupuis]

That test isn't touched by any changes made in this PR, so I say we ignore it.

[dyfer]

Thanks!

[dyfer]

I've updated lists of tests in #5359 and #5360