The image class has a strict internal representation so we dont need "format".
PixelFormat enum can be passed to functions that modify the pixel data (like setPixels()) so that a proper convertion can be done.
We don't want to expose any OpenGL functions to client code. Because if we do, there is a chance we break client code if we switch implementation (Direct3D).
When writing the X11 (linux) implementation there was a problem with X11 defining a "Display" type and we also have a Display class in the engine.
So to fix that problem and minimize the risk for running into other name conflicts. We move everything from global namespace.
Pass Renderer2D to Renderable and have each subclass of Renderable decide what the Renderer should do (like drawText(), drawRect() etc).
This makes the Renderable more flexible. right now, each renderable has a texture, vertices and indices. but what if some renderables does not use textures? or more than one? What if some renderables are just a group of other renderables? (Scene Graph's).
If we instead just make Renderables implement render(Renderer2D& r) interface. we can make each renderable pass the data it holds to the renderer without a hard defined interface.
