Function safe_arch::move_mask_m256

#[must_use]pub fn move_mask_m256(a: m256) -> i32

This is supported with target feature avx only.

Collects the sign bit of each lane into a 4-bit value.

assert_eq!(
  0b00110100,
  move_mask_m256(m256::from([1.0, 12.0, -1.0, 3.0, -1.0, -2.0, 3.0, 4.0]))
);