More or less in a working state - preliminary thesis version

I will probably use this version for my thesis (unless a big bug shows up unexpectedly).
The package works for my use case, though it could still be improved upon.
Some things that bother me still:

• For proper numba performance, the backend of the data should be moved from awkward arrays to numba List objects. Then, nopython mode can be used.
• The input/output is still very much oriented around my needs and could be more general.
• The deadlayer model is rudimentary (simple in/out of volume, w/o weighting).
• More examples.
• More tests.
• ...