Improved boolean definitions

[noodlecollie]

I've had a bit of a trip tracking down a completely bamboozling bug, where certain member offsets in my structs were inconsistent throughout my program when compiling on Windows. It turns out this was caused by the fact that raylib.h defines its own bool data type if it deduces that stdbool.h cannot be used, and the size of this type is not necessarily the same as the size of bool as stdbool.h defines it. Consequently, depending on whether the compiler had seen stdbool.h or raylib.h when it came across the definition of my structs, it would generate different member layouts in different compilation units for those structs which had bool members. If I passed one struct from one compilation unit to another, member values would appear to be completely different.

The reason why raylib deemed stdbool.h unavailable was because MSVC does not define __STDC__ unless the /Za compiler option is provided. Even after setting an explicit C standard in my CMake project, and disabling compiler extensions, CMake still did not add this compiler option. Rather than rely on that option, and only fix this problem for myself, I thought I'd make a PR to improve the way the boolean type is defined in raylib.

This PR makes the following changes:

• The preprocessor check regarding the availability of stdbool.h (at least, every check that I could find) has been updated to explicitly check the MSVC version if the __STDC__ check fails. Since Microsoft added C99 support in VS2013, the check has been augmented to check whether _MSC_VER is at least 1800.
• According to best practices, the definitions of the custom false and true types have been updated to false = 0 and true = !false. This ensures that false is always and obviously 0, and that true is defined as whatever the compiler considers not to be false, as opposed to a numerical 1.
• For some reason, raudio.c forcibly undefined the bool type on Windows, and this caused this file to fail compilation with the new changes. I'm not sure why that was the case, but removing this code fragment did not break compilation for me with the new changes added.

[raysan5]

@noodlecollie Wow! This is a good catch! I kind-of remember facing this same issue with bool size in the past but I can't remember exactly when and where... In any case, thank you very much for the improvement!