Function to_f32 

pub fn to_f32(h: u16) -> f32

Convert IEEE half-float bits to f32.

Every half-float value is exactly representable as an f32, so this conversion is lossless (including subnormals, signed zero, infinities and NaN).

assert_eq!(rolcb::float16::to_f32(0xC000), -2.0);
assert!(rolcb::float16::to_f32(0xFFFF).is_nan());