T23: Implement algorithms for thick maze generation via cellular automata

[sraaphorst]

I implemented maze-generating cellular automata using:

Neighbourhoods:

1. Moore

2. von Neumann extended

Strategies

1. Mazecetric

2. Maze

3. Vote45

4. Vote

They have various parameters, including max number of generations, max number of generations to store to detect configuration looping, probability for initialization for cells, neighbourhood functions, and behaviour functions.

Here is a ThickMaze using Moore neighbourhoods; the Maze behaviour function; 10,000 max generations; five stored generations; and 0.5 probability during initialization. As expected from cellular automata, it is neither connected nor perfect.