Function s15_fixed16_number_to_float 

pub(crate) const fn s15_fixed16_number_to_float(a: i32) -> f32

Produces the nearest float to a with a maximum error of 1/1024 which happens for large values like 0x40000040.