encoding: add compressor options in UseCompressor and SetSendCompressor

[Dostonlv]

Fixes #7017

RELEASE NOTES:

• encoding: Add optional compressorOptions to UseCompressor and SetSendCompressor, forwarded to Compress method.

[codecov]

Codecov Report

❌ Patch coverage is 57.69231% with 11 lines in your changes missing coverage. Please review.
✅ Project coverage is 82.25%. Comparing base (2b8b708) to head (6d65379).
⚠️ Report is 31 commits behind head on master.

Files with missing lines	Patch %	Lines
stream.go	55.55%	2 Missing and 2 partials ⚠️
rpc_util.go	57.14%	2 Missing and 1 partial ⚠️
server.go	60.00%	0 Missing and 2 partials ⚠️
encoding/gzip/gzip.go	0.00%	0 Missing and 1 partial ⚠️
internal/transport/server_stream.go	75.00%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #9036      +/-   ##
==========================================
- Coverage   83.01%   82.25%   -0.77%     
==========================================
  Files         413      413              
  Lines       33054    33315     +261     
==========================================
- Hits        27441    27403      -38     
- Misses       4205     4254      +49     
- Partials     1408     1658     +250     

Files with missing lines	Coverage Δ
encoding/encoding.go	68.96% <ø> (-6.90%)	⬇️
encoding/gzip/gzip.go	60.00% <0.00%> (-2.50%)	⬇️
internal/transport/server_stream.go	83.87% <75.00%> (-12.74%)	⬇️
server.go	82.03% <60.00%> (-0.93%)	⬇️
rpc_util.go	62.43% <57.14%> (-20.55%)	⬇️
stream.go	62.91% <55.55%> (-19.42%)	⬇️

... and 40 files with indirect coverage changes

🚀 New features to boost your workflow:

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

[Dostonlv]

@easwars assalamu alaykum could you review my code?

[Pranjali-2501]

Hi @Dostonlv , Thanks for raising the PR.

I have added a few comments, PTAL.

[Dostonlv]

@Pranjali-2501 Assalamu alaykum added tests to verify compressor options are correctly propagated. Please review.

[Pranjali-2501]

@arjan-bal, assigning it to you for second review.

[arjan-bal]

I left a few questions on the proposal: #7017 (comment)

I'll hold off on the PR review until we get confirmation on those points.

[arjan-bal]

Let's use the mutex to protect compressInvokes too. Unifying the synchronization mechanisms will make the code much simpler.

Additionally, please use empty lines to group the mutex-guarded fields, similar to this example:

grpc-go/internal/balancer/gracefulswitch/gracefulswitch.go

Lines 50 to 63 in 89fe783

	// mu protects the following fields and all fields within balancerCurrent
	// and balancerPending. mu does not need to be held when calling into the
	// child balancers, as all calls into these children happen only as a direct
	// result of a call into the gracefulSwitchBalancer, which are also
	// guaranteed to be synchronous. There is one exception: an UpdateState call
	// from a child balancer when current and pending are populated can lead to
	// calling Close() on the current. To prevent that racing with an
	// UpdateSubConnState from the channel, we hold currentMu during Close and
	// UpdateSubConnState calls.
	mu sync.Mutex
	balancerCurrent *balancerWrapper
	balancerPending *balancerWrapper
	closed bool // set to true when this balancer is closed

[arjan-bal]

style: The helper seems to be abstracting most of the test logic here. I would recommend avoiding the table driven approach and have separate tests for unary and streaming RPCs, inlining the helpers. Same for the TestUseCompressorOptionsPropagate test.

See https://google.github.io/styleguide/go/best-practices#leave-testing-to-the-test-function

[arjan-bal]

nit: You can use cmp.Diff to compare the entire slice.

[arjan-bal]

After discussing this with the other maintainers, we agreed that we need to provide a migration path for custom compressor implementers rather than breaking the API in a single release. We will need a transitional gRPC version where both the old and new APIs coexist before we fully remove the old one. I am currently evaluating the best approach to achieve this.

#8683