Documentation is needed in multiple places.

I've opted to use Sphinx to generate the documentation.
We should use numpy style docstrings: https://numpydoc.readthedocs.io/en/latest/format.html

Any functions that are supposed to be "public-facing" should have full docstrings (and probably doesn't hurt to have documentation for other internal functions).

Places that still need documentation:

• base.types
• base.kernel
• base.random
• base.util
• base.algorithms.mcmc
• base.algorithms.smc
• base.algorithms.smc_components
• base.algorithms.decision_forest
• base.executor
• domain.additive_structure
• domain.decision_tree
• domain.spectrum

Work started on docs branch