feat: add float types #17
Description
Add the following types:
f16f32f64f80f128
Every single one of these should be classified as a generic type and thus able to be used with shadow functions. These types also guarantee for CPU-instruction based handling rather than language-level handling.
Additionally, a special type can be made: f[m][e] which would allow to specify the amount of bits e and m. This however will not guarantee CPU-instruction based handling but will try to lower to the other float types for optimization. This type will also be experimental and probably not added until a long time in the future