MathF-functions are faster than the Math-counterpart and it looks cleaner, so use MathF when we cast to float or int anyway.
This was never batching, ever. Pointless memory overhead.
Depends on https://github.com/ppy/osu-framework/pull/837