Convert tests to JUnit5

[jonahgraham]

#1375 identified that some tests are not running in JUnit5 runner because discovery is different than JUnit3 runner.

This PR changes the tests to use JUnit5 (in core tests only for now). However much of the purpose of this PR is to modernize the tests. This PR also works around eclipse-jdt/eclipse.jdt.ui#2591

Fixes #1375

[jonahgraham]

There are more commits coming on this PR

[github-actions]

Test Results

  584 files   -   1    584 suites   - 1   13m 47s ⏱️ -1s
9 920 tests +  3  9 878 ✅  -  15  24 💤 ±0  14 ❌ +14  4 🔥 +4 
9 839 runs   - 116  9 797 ✅  - 134  24 💤 ±0  14 ❌ +14  4 🔥 +4 

For more details on these failures and errors, see this check.

Results for commit 863c40b. ± Comparison against base commit 6574af3.

This pull request removes 1331 and adds 1334 tests. Note that renamed tests count towards both.

NonIndexingTests ‑ test16bitUnicodeEscapeSequence
NonIndexingTests ‑ testAccessGlobalVariableFromConstexprFunction
NonIndexingTests ‑ testAccessGlobalVariableFromGlobalConstexpr
NonIndexingTests ‑ testAccessMemberOfCompositeParameter
NonIndexingTests ‑ testAccessOfStaticField
NonIndexingTests ‑ testAliasDeclaration
NonIndexingTests ‑ testAliasTemplate1
NonIndexingTests ‑ testArgumentEvaluation
NonIndexingTests ‑ testAssignmentOfArrays
NonIndexingTests ‑ testAssignmentOfCompositeValues
…

org.eclipse.cdt.core.parser.tests.ast2.cxx14.constexpr.ArrayTests$NonIndexingTests ‑ testAssignmentOfArrays
org.eclipse.cdt.core.parser.tests.ast2.cxx14.constexpr.ArrayTests$NonIndexingTests ‑ testAssignmentOfMultiDimensionalArrays
org.eclipse.cdt.core.parser.tests.ast2.cxx14.constexpr.ArrayTests$NonIndexingTests ‑ testGlobalArrayAccessValue
org.eclipse.cdt.core.parser.tests.ast2.cxx14.constexpr.ArrayTests$NonIndexingTests ‑ testInitializationOfArrays
org.eclipse.cdt.core.parser.tests.ast2.cxx14.constexpr.ArrayTests$NonIndexingTests ‑ testInitializationOfMultiDimensionalArrays
org.eclipse.cdt.core.parser.tests.ast2.cxx14.constexpr.ArrayTests$NonIndexingTests ‑ testPassArrayToFunctionAsPointerAndModifyCell
org.eclipse.cdt.core.parser.tests.ast2.cxx14.constexpr.ArrayTests$NonIndexingTests ‑ testPassMultiDimensionalArrayToFunctionAsPointerAndModifyCell
org.eclipse.cdt.core.parser.tests.ast2.cxx14.constexpr.ArrayTests$NonIndexingTests ‑ testPointerArithmeticsOnMultidimensionalArray
org.eclipse.cdt.core.parser.tests.ast2.cxx14.constexpr.ArrayTests$NonIndexingTests ‑ testPointerDereferencingOnArrayName
org.eclipse.cdt.core.parser.tests.ast2.cxx14.constexpr.ArrayTests$NonIndexingTests ‑ testPointerToArrayReturnedFromMemberFunction1
…

♻️ This comment has been updated with latest results.

[i-garrison]

at least for this test this change did not worked well:

• as far as I can see in my local run IndexBindingResolutionTestBase.setUpStrategy() fires and runs parser before IndexConceptTests.cxx20SetUp is fired so it's already too late to tell parser to recognize concepts
• could be that a few other tests are affected by similar issue

[jonahgraham]

Thank you for having a sneak peak - I think IndexConceptTest should do:

	@Override
	@BeforeEach
	protected void setUpStrategy() throws Exception {
		TestScannerProvider.sDefinedSymbols.put("__cpp_concepts", "201907L");
		super.setUpStrategy();
	}

instead of its own BeforeEach.

I misunderstood something in my first pass in refactoring and I will correct it before marking ready for review.

[jonahgraham]

BTW The most recent push does not resolve this yet.

[jonahgraham]

The problem I introduced is that some setUp methods were like this:

setUp() {
  super.setUp()
  localsetup codehere
}

And some were like this:

setUp() {
  localsetup codehere
  super.setUp()
}

And I didn't adequately pay attention to the difference in these two.

[jonahgraham]

The sizes of the diff can make it hard to see the important stuff. I think there should be some combination of -I command line options for git diff to ignore some stuff. What I am using is meld with these Text Filters:

• ^\t+@Test
• ^\t+public static TestSuite.*
• ^\t+return suite\(.*
• ^\t+}
• ^import.*

This makes the number of diff manageable and easy enough to review.

[jonahgraham]

I have decided to split this up into separate PRs as reviewing multiple commits in a single PR made it hard to see what was happening reliably.

I will label them all https://github.com/eclipse-cdt/cdt/issues?q=state%3Aopen%20label%3Ajunit5 to make the easier to follow.