Fix building examples on OSX

Base Audio Context System

Keep asking myself why I maintain this rendering path... -___-

now it works

Not tested yet

Depending on mesh data, it can be loaded and default vertex attribute
location points are set, including colors, tangents and texcoords2

Now GLFW3 windows/context creation works ok and a sample red rectangle
has been drawn using rlgl. Next step is working in tracking
position/orientation maths and try to get a simple 3d scene...

- Renamed some functions for consistency (default buffers)
- Removed mystrdup() function (implemented inline)
- Renamed TextFileRead() to ReadTextFile()

Now floating point is either on or off. This simplifies the use of 16bit
vs float.

updateAudioContext is almost done

Ease of use considered in api and channels are more convenient as
unsigned char type.

Removed BeginDrawingEx()
Added Begin2dMode() and End2dMode()

First stage of audio API update

Improve DXT-ETC1 support on HTML5

Hiding the struct from user should protect from accidentally modifying
the mix channel. This could cause serious errors down the road.