Update Nexus tests to use parallelsuite

[long-nt-tran]

What changed?

• Migrate the NexusTestBaseSuite to extend a lower-level *testcore.TestEnv instead of testcore.FunctionalTestBase, and rename NexusTestBaseSuite to a more aptly-named NexusTestEnv
• Migrated all tests that uses the base to parallelsuite which required edits of the following types:
  • Change calls like s.GetTestCluster() to env.GetTestCluster(), because now these things are provided by the NexusTestEnv environment rather than the suite as before
  • A lot of tests needed testcore.WithDedicatedCluster() for isolation, otherwise tests using capture to assert on metrics would get polluted
  • s.Eventually(...) (or some other s.Assertion(...) type calls) --> require.Eventually(t, ...), so the failures are reported at the right test rather than at suite level. Previously tests were not ran in parallel so this may not have been an issue
  • s.GetTestCluster().FrontendClient()... --> env.FrontendClient()...
  • Creating dedicated cluster env inside subtests instead of outside
  • Misc. changes -- I've commented inline

Why?

Migrate to the latest and greatest, and make clearer the division between what's provided by the suite vs. env. This PR also enables parallel runs of tests, which sped up test runtime (see bottom of PR for rudimentary benchmarks).

How did you test it?

• built
• run locally and tested manually
• covered by existing tests
• added new unit test(s)
• added new functional test(s)

Test runtimes

nexus_api_test.go
Before:

$ go test ./tests/ -run 'TestNexusApiTestSuite' -count=1
ok      go.temporal.io/server/tests     7.588s

After:

$ go test ./tests/ -run 'TestNexusApiTestSuite' -count=1
ok      go.temporal.io/server/tests     2.799s

---

nexus_api_validation_test.go
Before:

$ go test ./tests/ -run 'TestNexusAPIValidationTestSuite' -count=1
ok      go.temporal.io/server/tests     2.309s

After:

$ go test ./tests/ -run 'TestNexusAPIValidationTestSuite' -count=1
ok      go.temporal.io/server/tests     1.384s

---

nexus_workflow_test.go
Before:

$ go test ./tests/ -run 'TestNexusWorkflowTestSuite' -count=1
ok      go.temporal.io/server/tests     11.083s

After:

$ go test ./tests/ -run 'TestNexusWorkflowTestSuite' -count=1
ok      go.temporal.io/server/tests     3.709s

[long-nt-tran]

since we're creating dedicated env per subtest for isolated metric captures, the endpoint needs to be created inside the subtest itself, so just having the name as the field here so each subtest can do

env.createNexusEndpoint(testcore.RandomizeStr("test-endpoint"), testcore.RandomizeStr("task-queue")),

[stephanos]

I don't know if I missed it before, but this needs to be s.Run (otherwise it won't run in parallel). And that will likely require env to be once in every Run to make the metric snapshots deterministic.

[stephanos]

Same here as above.

[chaptersix]

is there a way to modify this config after a dedicated cluster has been allocated ( in the test)?

I commented this out and only a few tests failed.

[stephanos]

Note we had a previous convo around this here: #9768 (comment)

tl;dr we want to make this a default to avoid having to mark it as dedicated when this default setting works just fine as it's a global config - it should be overridable.

[long-nt-tran]

endpoint name needs to be created in the subtest after env is created, hence needing a factory fn here

[stephanos]

LGTM! Please address the remaining comments before you merge.