Add CodSpeed continuous performance benchmarks

[codspeed-hq]

Summary

• Adds CodSpeed integration for continuous performance tracking of the taskflow runtime benchmarks
• Creates google_benchmark-based wrappers for 4 benchmarks: fib, skynet, nqueens, and matmul
• Adds a GitHub Actions workflow (.github/workflows/codspeed.yml) that runs benchmarks on push to main, PRs, and manual dispatch
• Adds the CodSpeed badge to the README

What's included

Benchmark files (cpp/codspeed/)

New directory containing CodSpeed-compatible benchmark wrappers using codspeed-cpp, which provides a google_benchmark compatibility layer:

• bench_fib.cpp - Recursive fibonacci (fib(30))
• bench_skynet.cpp - Skynet 100M task spawn benchmark
• bench_nqueens.cpp - N-Queens solver (N=14)
• bench_matmul.cpp - Recursive matrix multiplication (512x512)
• CMakeLists.txt - Build configuration using FetchContent for codspeed-cpp and CPM for taskflow

The benchmark parameters are tuned to be smaller than the main comparison benchmarks to keep CI runtimes reasonable under CodSpeed's simulation mode.

CI workflow (.github/workflows/codspeed.yml)

• Triggers on push to main, pull requests, and manual dispatch
• Uses OIDC authentication (no token needed)
• Builds and runs all 4 benchmarks in simulation mode

Next steps

Once the PR is merged:

1. CodSpeed will start tracking performance on every push to main
2. PRs will receive automatic performance regression reports
3. Flame graphs and execution profiles will be available on the CodSpeed dashboard

[tzcnt]

Reimplement the benchmark wrapper for TooManyCooks, NOT Taskflow. Those files are in ./cpp/ToomanyCooks

[PyXiion]

Bruh

[codspeed-hq]

Congrats! CodSpeed is installed 🎉

🆕 4 new benchmarks were detected.

You will start to see performance impacts in the reports once the benchmarks are run from your default branch.

Detected benchmarks

• BM_NQueens: 19.7 s
• BM_Skynet: 39.1 s
• BM_Fib[30]: 389.9 ms
• BM_Matmul[512]: 404.1 ms

---