TextRender: Implement Z-buffer

[Makman2]

This feature is not downwards compatible since it uses another component of a vector that contains the depth-information. Projectionmatrices are now generated from OrthogonalDepthProjection (see #29).

[Makman2]

I think the only performant and viable solution is to implement a new renderer (TextDepthRenderer/TextZRenderer/similar), though this results in code duplication inside the existing TextRenderer, the drawing algorithms need tweaking everywhere for Z-capability, and if someone doesn't want it, it's not easy to turn it off/on without losing much performance while avoiding code-duplication.

[Makman2]

Creating a new renderer is really high code duplication, one can use a bool template argument zbuffer to specify different render behaviour. This means everything regarding render needs to be moved inside a .hpp.

[Makman2]

If somebody renders something, he/she will use in 99% of cases the z-buffer. Even for point rendering, a z-buffer that correctly controls what is rendered on top improves the render result. So it should be implemented by default. People have to look out to use the correct transformations.