todo: improve benchmarks

[CrabNejonas]

I performed some basic benchmarking and set up a rough harness for it, but it's not great yet. In particular it would be nice to:

1. Benchmark the wasm build, not a native one.
2. Benchmark the wasm build in-browser to get an idea of real-world perf.
3. Have it run in CI

We already have an in-browser testing setup using vitest, maybe we can reuse components from that 🤔