docs: add benchmark harness guide

[Karib0u]

Summary

• Add repeatable Windows and Linux benchmark harness scripts
• Add a public benchmarking guide with validity criteria, output files, and acceptance targets
• Add a standard-library corpus fetcher for reproducible benchmark rule sets
• Link the benchmarking guide from the docs index

Validation

• python -m py_compile scripts\rules\fetch_corpus.py
• PowerShell parser check for scripts\bench\windows.ps1
• bash -n scripts/bench/linux.sh via wsl.exe -d Ubuntu-26.04