Fix test server ignoring whole-second retry deadlines

[alvinsee]

TestServiceRetryState skipped the expiration check whenever the deadline's nanos field was zero. The guard appears to have been meant to detect an unset expiration, but the constructor already replaces an unset expiration with Timestamps.MAX_VALUE, which never compares as expired. As a result, the guard only suppressed real deadlines that happened to fall exactly on a second boundary, so retries were scheduled past their expiration instead of returning RETRY_STATE_TIMEOUT.

Drop the nanos guard and rely on the timestamp comparison alone. Add unit tests covering whole-second, sub-second, unset, and not-yet-reached expirations.

What was changed

Removed the expirationTime.getNanos() != 0 guard from the expiration check in TestServiceRetryState.getBackoffIntervalInSeconds, and added TestServiceRetryStateTest.

Why?

The guard only looked at the subsecond field of the protobuf Timestamp. An unset expiration is already turned into Timestamps.MAX_VALUE in the constructor, and that never compares as expired. So the guard never protected the "no deadline" case. Its only effect was to skip the timeout when a real deadline fell exactly on a whole second. In that case the test server kept scheduling retries past the deadline instead of returning RETRY_STATE_TIMEOUT.

Checklist

How was this tested:
The new test expirationOnWholeSecondTimesOut fails without the fix(expected:<RETRY_STATE_TIMEOUT> but was:<RETRY_STATE_IN_PROGRESS>) and passes with it. The full :temporal-test-server:test suite passes.

[CLAassistant]

All committers have signed the CLA.

[maciejdudko]

Hi @alvinsee, thank you for your contribution! The code looks good and the tests pass, I'm ready to approve it. But we just introduced a new process where every PR should add a note to the new CHANGELOG.md file. Please update the branch and add a note under Fixes section. You can look at other SDK repos for guidance, e.g. https://github.com/temporalio/sdk-dotnet/blob/main/CHANGELOG.md?plain=1

[alvinsee]

Thanks @maciejdudko! I rebased the branch onto the latest main and added an entry under Fixed in CHANGELOG.md (c49c7ec). Ready for another look.