benchmarks: add atomic one-shot builtin batch

[emil14]

Summary

This PR is the next slice after #1073 and adds the first truly atomic builtin runtime benchmarks.

Included

• benchmarks/atomic/builtin_pass
• benchmarks/atomic/builtin_lock

Both benchmarks use a single :start event and keep the measured builtin on the hot path without streams.Range, streams.Wait, or sync.WaitGroup in the measured flow.

Notes

• The previous fan_in and fan_out drafts were intentionally removed from this PR.
• Those components need additional support wiring to produce a valid standalone benchmark, and that methodology should be reviewed separately instead of being mixed into the first atomic batch.

Validation

• go test ./benchmarks -run=^$ -bench BenchmarkRuntimeE2E -benchtime=1x -count=1
• go run github.com/golangci/golangci-lint/v2/cmd/golangci-lint@v2.5.0 run ./benchmarks