internal/transport: Wait for server goroutines to exit during shutdown in test

[arjan-bal]

Internal bug: b/416700146

Presently the PingPong tests don't wait for server goroutines to finish work before the return. This causes data races when the next test tries to update the GRPCLogger. Closing the server isn't sufficient.

grpc-go/internal/transport/http2_server.go

Lines 1265 to 1268 in 950a7cf

	// Close starts shutting down the http2Server transport.
	// TODO(zhaoq): Now the destruction is not blocked on any pending streams. This
	// could cause some resource issue. Revisit this later.
	func (t *http2Server) Close(err error) {

RELEASE NOTES: N/A

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 82.35%. Comparing base (515f377) to head (6c43100).
Report is 21 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #8306      +/-   ##
==========================================
+ Coverage   82.14%   82.35%   +0.20%     
==========================================
  Files         417      419       +2     
  Lines       41344    41988     +644     
==========================================
+ Hits        33961    34578     +617     
- Misses       5957     5963       +6     
- Partials     1426     1447      +21     

see 50 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]

Should this (and similar) be outside the other goroutine now? Or does HandleStreams track this and not return until all streams it's handling are done?