Function atan

pub fn atan(x: f32) -> f32

Compute a fast approximation of the arctangent of x.

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

See also atan_raw which only works on |x| <= 1, but is faster.