* Lua: All usage of double converted to float. printf("%f") still uses promotion to double, (floating point va_args are always promoted to double)
Speed improvements:
* reordering of rotable to put most used tables at the top
* caching of last found global table
Use real pointer to rotable instead of misusing void* to pass integer value
Lua internal traces now turned on by TRACE_LUA_INTERNALS cmake option
* Replaced floating point division with multiplication (where possible), the hardware FPU is much faster at multiplication vs division.
stb_image: enabled STBI_NO_HDR and STBI_NO_LINEAR (removes code that we don't need)
