Add color/depth write mask support (and rename GPU_SetDepthTest() accordingly).
Add blending color.
Blending color must be set right after stencil op. GPU freezes otherwise.
* add GPU_FinishDrawing() to be called after a batch of GPU_DrawArray() calls if you're done drawing or if you intend to change the GPU configuration before drawing more. Also fix GPU_Finalize(). With those changes, the GPU no longer freezes if you call GPU_DrawArray() an even number of times.
* fix GPU_SetViewport() to allow color buffer reading, so blending and logicop work as expected.
