ci: improve flaky tests

[vaind]

Some tests are flaky if the machine is overloaded - some goroutines may not get captured consistently on all ticks. And since we cannot control the number of goroutines running (HTTP workers from other tests hang around indefinitely, see #731). Disabling the flaky test in CI when race detection is on. We can enable it once #731 is fixed.

The other flaky test I'm not sure about so just adding logs and maybe it would hep figure out what's wrong:

--- FAIL: TestProfilerPanicOnTick (0.00s)
    profiler_test.go:339: 
        	Error Trace:	/home/runner/work/sentry-go/sentry-go/profiler_test.go:339
        	            				/home/runner/work/sentry-go/sentry-go/profiler_test.go:263
        	Error:      	Should NOT be empty, but was 
        	Test:       	TestProfilerPanicOnTick

#skip-changelog

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Files	Coverage Δ
internal/testutils/race.go	100.00% <100.00%> (ø)

📢 Thoughts on this report? Let us know!.

[greywolve]

LGTM