It seems now it's required to use archiver `llvm-ar`

* Fixed gltf missing transforms on load

mend

* extracted Matrix calculation in to static method and added skinning check

* fixed formatting

* new models_magicavoxel_loading example

* Portable header-only file "magicavoxel_loader.h" for MagicaVoxel loader example.

* models_magicavoxel_loading example added to CMakeLists.txt and Makefile

* fix models_magicavoxel_loading example for linux.

* * vox_loader into "src/external/vox_loader.h"
 * vox file support for "models.c"
 * updated example "models/models_magicavoxel_loading.c"

* * Fix Vox_FreeArrays (removed memory leak)

* * removed magicavoxel_loader.h

* * Revert vs2019 solution

RENAMED: GetCodepointsCount() -> GetCodepointCount()
RENAMED: GetTouchPointsCount() -> GetTouchPointCount()

Following english rules, it should be singular name before Count.

Note this new raygui version embeds ricons and `GuiTextBoxMulti()` is not working properly

* Fix tcc not finding emmintrin.h

This allows to compile raylib with tcc with no errors.

* Remove __TINYC__ check from stb_image.h

This will be placed under textures.c

* Move tcc check to textures.c

Avoiding to change stb_image.h

As far as next raylib version will be 4.0, no backward compatibility hacks will be maintained.

This is a big change for optimization and a more professional understanding of audio. Instead of dealing with samples, now we deal with frames, like miniaudio does, so, avoiding continuous conversions from samples to frames.

Several breaking changes have been done lately so I think it's better to mark raylib for next release as 4.0.

Note that `extern "C"` calling convention only affects objects that need to be seen by the linker, in our case only functions... but it would also be required by global variables exposed, if any.

* GenMeshTangents Fix

* GenMeshTangents Fix - Comment Update

* GenMeshTangents Fix - Comment Update final

* Code Style Changes

* Code Style Changes Final

* Code Style Changes Final 2

* GenMeshTangents better handling for issue #1876

* GenMeshTangents better handling for issue #1876

* GenMeshTangents: Better fix for issue #1876

* vboId location fix

* rlUpdateVertexBuffer - bufferId is an unsigned int

* Fix for issue#1809 - Obj with more materials then mesh caused crash

* Fix for issue#1809 - Formatting Update

- ADDED: rlEnableColorBlend()/rlDisableColorBlend()

* GenMeshTangents Fix

* GenMeshTangents Fix - Comment Update

* GenMeshTangents Fix - Comment Update final

* Code Style Changes

* Code Style Changes Final

* Code Style Changes Final 2

* GenMeshTangents better handling for issue #1876

* GenMeshTangents better handling for issue #1876

* GenMeshTangents: Better fix for issue #1876

* vboId location fix

* rlUpdateVertexBuffer - bufferId is an unsigned int

This commit introduces some breaking changes for library consistency, hopefully not too dramatic... Here the full list:
 - RENAMED: struct `CharInfo` to `GlyphInfo`, actually that's the correct naming for the data contained. It contains the character glyph metrics and the glyph image; in the past it also contained rectangle within the font atlas but that data has been moved to `Font` struct directly, so, `GlyphInfo` is a more correct name.
 - RENAMED: `CodepointToUtf8()` to `CodepointToUTF8()`, capitalization of UTF-8 is the correct form, it would also require de hyphen but it can be omitted in this case.
 - RENAMED: `TextToUtf8()` to `TextCodepointsToUTF8` for consistency and more detail on the functionality.
 - ADDED: `GetGlyphInfo()` to get glyph info for a specific codepoint
 - ADDED: `GetGlyphAtlasRec()` to get glyph rectangle within the generated font atlas
 - Some additional tweaks for consistency

WARNING: DrawTextPro() requires including `rlgl.h`, before it was only dependant on `textures` module.

- DrawTextRec() and DrawTextRecEx() have been moved to example, those functions could be very specific depending on user needs so it's better to give the user the full source in case of special requirements instead of allowing a function with +10 input parameters.

I think it is useful to know what the empty behavior is in cases where
you are enumerating key presses (e.g. implementing "press any key to
continue").