add benchmark to CI/CD

[agavra]

fixes #29 by running benchmarks on CI/CD and reporting the changes on the pull request - since both baseline and new run on the same GHA runner the diff should be somewhat reliable even if the absolute value of the numbers are not.

[github-actions]

📊 Benchmark Comparison Report

This pull request includes Criterion benchmarks comparing performance to the main branch.
This comment will automatically update as the benchmarks are re-run on each commit.

The table below shows relative ratios and timing stats for each benchmark group:

group                main                                   pr
-----                ----                                   --
ops/merge            1.03  1509.2±26.32ns        ? ?/sec    1.00  1463.8±18.04ns        ? ?/sec
ops/project          1.00    566.6±8.51ns        ? ?/sec    1.05    594.0±7.81ns        ? ?/sec
serde/deserialize    1.00    314.7±4.43ns        ? ?/sec    1.03    324.0±4.70ns        ? ?/sec
serde/serialize      1.00    310.8±7.61ns        ? ?/sec    1.00    310.5±5.01ns        ? ?/sec

✅ Benchmarks completed successfully.

🧠 Notes:

• These benchmarks are not a pass/fail gate and are informative only.
• Use this as a signal to review performance-sensitive changes.
• Results may be unreliable due to GHA runner hardware variance.
• If results indicate a significant performance regression, run the benchmarks locally to confirm.

Reported by the benchmark CI bot