Multiple integration test suites have flaky failures due to race conditions, missing CRD optimization, and nil-safety issues in test teardown.

Global Hub 1.7.0

Yes - intermittent failures in CI, especially under load.

Run integration tests repeatedly:

make integration\-test

1. agent/status tests fail due to race condition on map access (concurrent goroutine writes to map[string]*cloudevents.Event while tests read)
2. operator/controllers transporter test fails with 409 Conflict (concurrent Kafka CR modification)
3. envtest suites fail with CRD install timeouts when loading all 35+ CRDs unnecessarily
4. AfterSuite panics on nil testPostgres when BeforeSuite fails

All integration tests pass reliably without flaky failures.

Fix PR: https://github.com/stolostron/multicluster-global-hub/pull/2286

Changes:
- Replace map with sync.Map for concurrent event access in agent/status tests
- Wrap transporter EnsureKafka in Eventually retry block for 409 conflicts
- Optimize each test suite to load only required CRDs instead of full directory (35+ CRDs)
- Add nil guards for testPostgres in AfterSuite across 4 operator/manager suites
- Remove StartEnvTestWithRetry utility in favor of CRD optimization approach

Generated with [Claude Code|https://claude.ai/claude-code]