Add reproducible benchmark suite and baseline results #15
Description
Summary
Add a first-class benchmark workflow so performance claims can be measured and tracked over time.
Proposed work
- Add a dedicated benchmark namespace with representative read/write scenarios.
- Add a Clojure alias and/or bb task to run benchmarks consistently.
- Capture baseline results in a .
- Document environment + run command so results are reproducible.
Why
Current benchmarking is ad-hoc in and is not easy to rerun or compare across commits.