Improved Handling Of Endianess Conversion When Using Physfs

[stefandee]

Currently converting little endian (which all the data files are in) to platform endian, is done through the PPT_BIGENDIAN_IO define in Stream_PhysFS

This is a bit awkward as you need to make specific builds for big endian machines.

Replace the define using physfs specific endianess conversion functions for integers and figure out a way to convert float/double.