codec/pcm Convert functions to natively calculate using bytes

[David Sutton]

Currently, all filters are contained as float slices, and are applied to a float slice. This requires the input data to be converted to floats, then back to bytes. Whilst the calculations don’t seem to be computationally intensive, it could be beneficial to increase efficiency, as well as simplify the code. The change could also minimize the risk of artifacting from a bad conversion. The change would simplify the code substantially, significantly shortening the length of code, as well as decreasing the function count (non-exported).