311 timeouts

[bretambrose]

• Adds support and tests for 311-client-wide operation timeout for Subscribe and all of its variants. The previous timeout implementation did not apply to subscribe.
• Adds internal APIs (and tests) for per-311-operation timeouts for Subscribe, Unsubscribe, and Publish

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (d55553f) 82.72% compared to head (3e27f40) 82.77%.

Additional details and impacted files

@@               Coverage Diff               @@
##           311Listener     #347      +/-   ##
===============================================
+ Coverage        82.72%   82.77%   +0.05%     
===============================================
  Files               23       23              
  Lines             9088     9115      +27     
===============================================
+ Hits              7518     7545      +27     
  Misses            1570     1570              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[sbSteveK]

It looks like we're using the same value for the regular timeout tests and overridden timeout tests (3 seconds). Let's use different values (shorter for override) to make sure we're using the override value when examining test results. Also, might be useful to set the protocol_operation_timeout_ms in the override tests to make sure we're overriding the provided value explicitly. Not setting it defaults the regular value to UINT64_MAX which is ignored so any timeout occurring probably indicates this test is successful.

[bretambrose]

I'm not bothered by this. Elapsed time comparisons in tests have consisently blown up in our faces when containers and CPU throttling on cloud CI (and emulated ARM builds) get thrown into the mix.

[sbSteveK]

Some trivial comments that may or may not need action but otherwise looks good.