Grimme D3-dispersion energy and their derivatives

This is an temporary experimental side-repo derivative work of this work: https://github.com/Rafael-G-C/pyDFTD3
Which is a derivative work of https://github.com/bobbypaton/pyDFTD3, (c) 2020 Rob Paton and contributors
Which is a Python reimplementation of the Grimme D3-dispersion energy

Notes about this repository

The code can and probably will be sent as pull request towards https://github.com/Rafael-G-C/pyDFTD3 but I didn't want to undo a lot of other people's work with one PR without asking so this requires some discussion first.
For simplicity when refactoring I have removed the Becke-Johnson damping scheme but it can be reintroduced.
I have moved out all parameter and I/O out of the d3 function so that it does not have to know anything about functionals and setting the appropriate parameters can be done caller-side.
It turned out to be significantly more efficient to compute all derivatives of one order in one shot. demo.py shows how this can be done.
Distance-based screening can be done but is not completely trivial and hasn't been attempted in this rewrite.