Use macros instead of inline for the generic (non-optimised) float->int