Function safe_arch::reciprocal_sqrt_m128_s

#[must_use]pub fn reciprocal_sqrt_m128_s(a: m128) -> m128

This is supported with target feature sse only.

Low lane 1.0 / sqrt(a) approximation, other lanes unchanged.

let a = m128::from_array([16.0, 8.0, 9.0, 10.0]);
let b = reciprocal_sqrt_m128_s(a).to_array();
let expected = [0.25, 8.0, 9.0, 10.0];
for i in 0..4 {
  assert!((b[i] - expected[i]).abs() < 0.001);
}