docs(benchmark): add throughput and latency benchmarking protocol (#106)

[showmemercy19-rgb]

Adds throughput and latency benchmarking protocol.

[github-actions]

All applicable PR CI workflows passed for 5b9feda.

• ci
• security

Ready for maintainer code review.

[armstrongsam25]

Thanks for contributing. I reviewed the full diff against current main and the LETS constitution.

The complete benchmarking protocol is two bullets and supplies no workload definition, exact source/configuration, procedure or retained measurements for the stated 10k req/s P99 scenario. Constitution XI prohibits performance/scale claims exceeding retained evidence. Please provide a reproducible benchmark procedure with explicit measurement-versus-target labels, durability posture, workload/cost vectors, output/evidence retention and existing harness references; do not imply 10k req/s is qualified capacity.

Please expand this into complete, source-backed documentation before merge (constitution XI and XII). This head contains no measurable executable lines, so changed-line coverage is not applicable; that should be recorded explicitly in the PR. No tests or candidate code were executed during this review.

[armstrongsam25]

Closing after source-based triage of head 5b9feda. The proposed benchmark protocol only names a load/latency scenario and contention profiling. It contains no runnable command, workload definition, configuration, measurement procedure or output/evidence retention instructions. It contributes no reproducible protocol; this decision does not assume that the mentioned load was asserted as measured capacity.

Evidence: existing benchmark/evidence routing; evidence requirements; entire submitted benchmark protocol.

Triage criterion: unsupported-completion. Closing this unmerged PR does not close an unfulfilled task, delete the contributor branch, or award bounty points. A distinct substantive revision may be reviewed after reopening.