Speed up local Go test runs

[bigboateng]

Summary

• add a CPU-aware local runner that verifies an exact test partition and runs isolated serial workers
• make the faster runner the documented local default while keeping serial Unix CI unchanged
• add the required release note

Test Plan

• python3 -m unittest discover -s .github/tests -p "test_*.py" -v
• python3 .github/scripts/test_ci_shard.py
• python3 .github/scripts/run_go_tests.py — 743 tests passed in 80.1s with 10 workers
• go -C boatstack vet ./...
• go -C boatstack build ./...
• release-note validation and git diff --check
• exact-head serial macOS and Ubuntu CI plus all six Windows shards