fix: remove avx2 runtime detection

[DaniPopes]

Small overhead due to necessary function call, and at that point the non-avx2 path is similar in performance. In practice this doesn't matter if you compile with -Ctarget-cpu=native or similar since most modern CPUs have AVX2

[codspeed-hq]

Merging this PR will improve performance by 50.15%

⚡ 24 improved benchmarks
✅ 82 untouched benchmarks
⏩ 111 skipped benchmarks¹

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
⚡	le/8	794.7 µs	563.8 µs	+40.95%
⚡	le/64	794.7 µs	563.8 µs	+40.95%
⚡	cmp/16	791.9 µs	561 µs	+41.15%
⚡	le/32	794.7 µs	563.8 µs	+40.95%
⚡	lt/64	794.7 µs	563.8 µs	+40.96%
⚡	lt/8	794.7 µs	563.8 µs	+40.96%
⚡	lt/16	794.7 µs	563.8 µs	+40.96%
⚡	common_prefix_length/32	744.5 µs	496 µs	+50.12%
⚡	lt/32	794.7 µs	563.8 µs	+40.96%
⚡	cmp/32	791.9 µs	561 µs	+41.15%
⚡	cmp/64	791.9 µs	561 µs	+41.15%
⚡	cmp/8	791.9 µs	561 µs	+41.15%
⚡	common_prefix_length/64	744.6 µs	542.8 µs	+37.17%
⚡	le/16	794.7 µs	563.8 µs	+40.95%
⚡	common_prefix_length/8	744.6 µs	495.9 µs	+50.15%
⚡	common_prefix_length/16	744.5 µs	496.1 µs	+50.06%
⚡	gt/8	794.7 µs	563.8 µs	+40.94%
⚡	gt/64	794.7 µs	563.8 µs	+40.94%
⚡	ge/16	797.4 µs	563.8 µs	+41.44%
⚡	ge/8	797.4 µs	563.8 µs	+41.44%
...	...	...	...	...

ℹ️ Only the first 20 benchmarks are displayed. Go to the app to view all benchmarks.

---

_{Comparing dani/x86 (0e9c616) with main (89d5790)}

Footnotes

1. 111 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩