This commit fixes the bug that `glnvg__renderFill()` allocates unused quad vertices for `GLNVG_CONVEXFILL`, which causes unused vertices uploaded to the GPU.

Besides, the quad for the `GLNVG_FILL` type is changed to drawn by `GL_TRIANGLE_STRIP` instead of `GL_TRIANGLES` so two more vertices can be reduced per quad.

In one of my test case, I need to paste an image to a framebuffer about 5,000 times, and this change would reduce the total number of uploaded vertices from `52,106` to `22,476`. It is good for FPS and battery consumption.

This commit moves the `discard` instruction in the fragment shader to an earlier position to avoid performing calculations whose results are unused.

This commit fixes the issue caused by #357 that nanovg failed to compile on iOS due to the wrong way of including glext header.

To make the examples link correctly under OS X 12.3, you need to include
<OpenGL/glext.h> in the file.  This commit limits what happens by protecting
the include as follows:

    #ifdef __APPLE__
    #   include <OpenGL/glext.h>
    #endif

Thus, this should have no effect on non-Apple systems.

This commit fixes issue #167 that `nvgImagePattern()` does not handle coordinate well if the specified image applies the `NVG_IMAGE_FLIPY` parameter.

There was no nvglImageFromHandleGL2 implementation, it must have been
an error during some refactoring

This commit fixes #300 and #306 that “control may reach end of non-void function” error occurs in `glnvg_convertBlendFuncFactor()`. The function now returns `GL_INVALID_ENUM` for invalid blend factor and if that happens, `glnvg__blendCompositeOperation()` uses default source-over composite operation.

This commit updates the implmentation of global composite operation to follow the advices described at c4b865796d (r74248556)

This commit implements the `nvgGlobalCompositeOperation()` function to support blending between frames. All operations defined in HTML5 canvas API are supported. Also, it is possible to create custom composite operation by calling `nvgBlendFunc()` or `nvgBlendFuncSeparate()` functions.

This commit updates the `nvgBeginFrame()` function to forward the received `devicePixelRatio` parameter to the `renderViewport()` function so it is possible to configure the environemnt correctly when using `bgfx` backend.

in some situations int and GLsizei is not the same type which leads to conversion warnings

The syntax is wrong when checking defined identifier.

This commit implements a new nvgCancelFrame() function for the purpose described in #168.

Currently, if calling nvgBeginFrame(), the nvgEndFrame() function must be called to finish the drawing. However, if knowing the drawing would take too much time and want to give up the drawing. There is no way to cancel it. nvgCancelFrame() gives another choice to end the frame without actually drawing.

Conflicts:
	src/nanovg_gl.h

- API changed!
- moved image repeat from nvgImagePattern() into image creation
nvgCreateImage*()
- made flip-y and premultiplied common image flags (not just GL)
- removed nvglImageFlags(), flags passed in via
nvglCreateImageFromHandle() flags
- nvgluCreateFramebuffer takes image flags as param

- added NVG_DEBUG which can be used to omit calls to glGetError()
- removed return value for glnvg__checkError
- create flag defines to enum

These changes make it possible to integrate nanoVG into other openGL libraries like Cinder and openFrameworks and be able to draw afterwards.

- added typedefs for all structs
- fixed some compiler warnings

Tweaked premake to remove some windows warnings.

- all colors in the shader is handled as premultiplied
- removed alphaBlend param from nvgBeginFrame(), it became obsolete
- added NVGL_TEXTURE_PREMULTIPLIED flag
- separated FBO example
- FBOs are treated as premultipied for correct alpha
- changed FBO storage from depth/stencil 24/8 to stencil 8
- moved discard to end of shader which restores the render speed again

- added (optional) 3-pass stencil trick to stroke rendering, reduces
overdraw artifacts, a little slower
- fixed some text bounds issues

- Added alpha for images
- Added global alpha

- fixed flags setup in nvglCreateImageFromHandle
- changed frame buffer to only compile on GL3
- changed nvgluCreateFramebuffer() to return FBO struct instead of
filling existing one (for consistency)

- use regular uniforms for once-per-frame values
- set tex sampler when using NANOVG_GL_USE_UNIFORMBUFFER

"Fragment Shader is going to be recompiled because the shader key based on GL state mismatches."