test: remove hardcoded queue integration timeout and parallelize workers

[sbalabanov]

Improve Integration Test Runtime
Remove hardcoded timeout in lieu of Bazel's timeout

Summary

• coordinate concurrent MySQL message-queue test receivers through one shared result channel
• keep message aggregation and duplicate tracking exclusively on the main test goroutine, without mutexes or a wait group
• acknowledge each delivery before reporting it, cancel on successful completion or the first real worker error, and consume every worker's terminal result before returning
• distinguish expected context cancellation after completion/error from standalone cancellation that must fail the test
• remove local integration-test safety timeouts and rely on Bazel's test timeout for stuck tests

Benchmark

Uncached Bazel test attempt runtime:

• before: 58.3s
• after: 38.3s
• improvement: 20.0s (~34%)

Tests

• ./tool/bazel test //test/integration/extension/messagequeue/mysql:go_default_test --nocache_test_results --test_output=errors --strategy=TestRunner=local ...
• ./tool/bazel test //test/integration/extension/messagequeue/mysql:go_default_test --nocache_test_results --test_output=errors --strategy=TestRunner=local --@rules_go//go/config:race ...
• make fmt
• make lint
• make check-tidy
• make check-gazelle

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}