Circle Collision

Did some collision detection algorithms for 2D circles and will extend them to 3D spheres. The algorithm uses a hill-climbing algorithm and thus usually misses the 'true' optimal solution.

Tried combining the grammar system I did last week with my viewer code from ages ago. Still works remarkable quickly. Next step, I'm going to look at some of Wittkower/Rowe's analysi and see what are the smallest number of functions needed to create 'parti' models of Palladio's plans.