In case mesh loading fails, a cube is generated instead!

Seems tcc doesn't provide __sync_*. See #435.

Reported in #435, tcc generates out-of-line definitions for inline
symbols, something it shouldn't.  This fix punishes tcc for that
by making applications it compiles (slightly) larger...

Added raylib version to config

I would have liked config.h to be selected by include dir configuration,
but this way is less intrusive.

That way, a user needs only to touch a single file to configure what
features raylib is built with.
Include guards are left out intentionally, because config.h should only
be included in source files, not headers.

Later on, config.h can also define the raylib version (#461).

Now PLATFORM checks only used on core and utils modules

Renamed SaveImageAs() to ExportImage() for consistency on what actually happens with data.

Not sure if math is ok... just left a commented piece of code that uses pre-multiplied alpha.

Reviewed for PLATFORM_ANDROID and PLATFORM_UWP

This is already the order that is used for Android. It doesn't appear to
make a difference on desktop but on web using the timer before it's been
initialized (by glfwInit, inside InitGraphicsDevice) causes the a long
(and variable but often several seconds) sleep between the first and
second frame.

Fixes: 468309d ("Early-exit InitWindow if InitGraphicsDevice fails")

Changes motivated by commentary in pull request 513

...PLATFORM_ANDROID on Windows, using MinGW-w64 (x86) provided Make (GCC
7.2)

Inspired by #504.
Instead of requiring the user to do PLATFORM_ANDROID #ifdefery,
have the android_main entry point exported by raylib and call
the user-defined main. This way many games could (in theory)
run unmodified on Android and elsewhere.

This is untested!

- SetWindowSize() to scale Windows in runtime
- SetMouseScale() to scale mouse input, useful when rendering game to a
RenderTexture2D that will be scaled to Window size (used on rFXGen tool)

GLFW has been updated to latest version, probably this code is not
required any more due to already been integrated into library... but it
needs to be tested...

Renamed function for consistency with a possible Mesh manipulation functions (maybe added in a future). Naming follows Image*() manipulation functions.

It has no sense to be inside LoadOBJ(), mesh processing moved to own functions: MeshTangents() and MeshBinormals(). Not exposed to user yet.