Base Performance Benchmarks

[austinorr]

this refactored bench suite is not meant to be pushed upstream. It's just here for reference by the upstream maintainers.

Requires x86 arch with support for sse2, sse41, and avx2 or will panic.

Workflow i used for benching other PRs:

1. add to a branch based on the release commit of simdeez v2.0.0.
2. capture a baseline for v2 with cargo bench --bench bench_main -- "float_rounding|int_ops|horizontal_add" --save-baseline baseline
3. add each linked PR in turn and rerun benches with cargo bench --bench bench_main -- "float_rounding|int_ops|horizontal_add" --baseline baseline to see incremental performance improvement

[austinorr]

The clippy-arm64 check fails because these benches work only on x86. Please disregard the failing check.