🐰 add codspeed benchmarking to ci/cd

[jvdd]

To make codspeed work I had to add "half" as default feature. I will revert this once CodSpeedHQ/codspeed-rust#1 is resolved

[codspeed-hq]

CodSpeed Performance Report

Merging #15 codspeed (4807792) will create unknown performance changes.

Congrats! CodSpeed is installed 🎉

🆕 44 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

• scalar_random_long_f32 (1.9 ms)
• sse_random_long_f32 (883.6 µs)
• avx_random_long_f32 (403.8 µs)
• impl_random_long_f32 (404 µs)
• scalar_random_long_i16 (2.1 ms)
• sse_random_long_i16 (411.4 µs)
• avx2_random_long_i16 (203.2 µs)
• impl_random_long_i16 (203.4 µs)
• scalar_random_long_i32 (2.1 ms)
• sse_random_long_i32 (819.4 µs)
• avx2_random_long_i32 (403.7 µs)
• impl_random_long_i32 (403.9 µs)
• scalar_random_long_f64 (2.1 ms)
• sse_random_long_f64 (1.8 ms)
• avx_random_long_f64 (805.3 µs)
• impl_random_long_f64 (805.5 µs)
• scalar_random_long_i64 (2.2 ms)
• sse_random_long_i64 (2 ms)
• avx2_random_long_i64 (869.2 µs)
• impl_random_long_i64 (869.4 µs)
• scalar_random_long_u32 (2.1 ms)
• sse_random_long_u32 (862.1 µs)
• avx2_random_long_u32 (403.7 µs)
• impl_random_long_u32 (403.9 µs)
• scalar_random_long_u64 (2.2 ms)
• sse_random_long_u64 (2.1 ms)
• avx2_random_long_u64 (869.3 µs)
• impl_random_long_u64 (869.5 µs)
• scalar_random_long_f16 (2.6 ms)
• sse_random_long_f16 (497.7 µs)
• avx2_random_long_f16 (236.3 µs)
• impl_random_long_f16 (236.5 µs)
• scalar_random_long_u16 (2.1 ms)
• sse_random_long_u16 (433.1 µs)
• avx2_random_long_u16 (203.3 µs)
• impl_random_long_u16 (203.5 µs)
• scalar_random_long_i8 (2 ms)
• sse_random_long_i8 (276.4 µs)
• avx2_random_long_i8 (200.8 µs)
• impl_random_long_i8 (276.6 µs)
• scalar_random_long_u8 (2 ms)
• sse_random_long_u8 (287.1 µs)
• avx2_random_long_u8 (197.4 µs)
• impl_random_long_u8 (287.2 µs)

[jvdd]

LGTM