Function fast_math::atan2

pub fn atan2(y: f32, x: f32) -> f32

Compute a fast approximation of the four quadrant arctangent of y and x.

The maximum absolute error across all f32s is less than 0.0038.