add throughput reporting, warmup, stats, and CSV export to exp_hash

[alekseevpavel04]

Summary

The hash benchmark only reported min time per run, making it hard to compare algorithms by actual throughput or spot variance across runs. Added throughput (MB/s) to the default output, a --stats flag for mean/stdev columns, a --warmup arg to discard cold-cache runs before timing, and --output to dump a CSV for further analysis.

All timing logic is now in _run_benchmark(), which also fixes a shadowing bug where the inner closure was named hash, overriding the builtin. A _print_summary() table at the end shows peak throughput per algorithm across all sizes, which is the number most people actually care about.

Run with --stats --output results.csv to get the full picture. Default behavior (just min time + throughput column) is backward compatible.

Checklist

• All commits are signed-off, using DCO
• All new code has docstrings and type annotations
• All new code is covered by tests. Aim for at least 90% coverage. CI is configured to highlight lines not covered by tests.
• Public facing changes are paired with documentation changes
• Release note has been added to CHANGELOG.md if needed

[mihaimaruseac]

You need to do the DCO for the commit.