Function safe_arch::sqrt_m256d

#[must_use]pub fn sqrt_m256d(a: m256d) -> m256d

This is supported with target feature avx only.

Lanewise sqrt on f64 lanes.

let a = m256d::from_array([1.0, 4.0, 9.0, 16.0]);
let b = sqrt_m256d(a).to_array();
assert_eq!(b, [1.0, 2.0, 3.0, 4.0]);