grpcsync/event: Simplify synchronization

[arjan-bal]

This change avoids using sync.Once and reuses the existing atomic boolean to ensure an event is fired only once.

RELEASE NOTES: N/A

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 82.31%. Comparing base (515f377) to head (07f1ca0).
Report is 26 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #8308      +/-   ##
==========================================
+ Coverage   82.14%   82.31%   +0.17%     
==========================================
  Files         417      419       +2     
  Lines       41344    41990     +646     
==========================================
+ Hits        33961    34565     +604     
- Misses       5957     5978      +21     
- Partials     1426     1447      +21     

Files with missing lines	Coverage Δ
internal/grpcsync/event.go	100.00% <100.00%> (ø)

... and 51 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[dfawley]

There is a subtle behavior change in that Fire can now return before the channel is closed. E.g.:

go e.Fire()  // Pretend this wins, but blocks for a second before close(e.c) happens
e.Fire() // This loses and returns false
select {
	case <-e.Done(): // This channel is still open
	default: panic("used to be impossible")
}

But, I don't think this is a behavior that is (or should be) relied upon, so it should be fine.