Arc-alkali-rydberg-calculator

Minor release fixes selection of basis in pair-state basis for calculations of very low states (like first excited state).
See 4e1ddbbe3075e3315469153a29b871667cb01d31 and d08ea621db8b9922f099223a195998dce86e4c92 .

Updated literature values for Caesium with values in Phys. Rev. A 94, 012505 (2016).

**For updating existing ARC version:**

* if you want to **keep your calculated dme** you have to manually copy `/arc/data/caesium_literature_dme.csv` into `~/.arc-data` on your computer;
* if you do **not want to keep your calculated dme** you can delete `~/.arc-data` folder, and on first loading of new folder will be created and populated.

Minor fix: getStateLifetime to include decays to levels that are for smaller
principal quantum number (n) than ground level, but are above in energy
due to angular part (see 5a73af2c27d0bee90ad8007a66d432b8a36ef28b)

This release does not introduce any changes to the code base.

Just forces wheels (distributed in installation with pip install...) to be build with old Numpy to resolve installation issues reported in 25 .

Minor release, prompted by (24):

* fixes problem that caused sometimes getC6pertubatively to crash (acaa32fe3f8ba8e21a0d0bc76ad4be9d3603e652)
* adds more choice for state highlighting (faaaa5fc338ef1aa63db2bcf0b31fead5c15e46b and [doc](https://arc-alkali-rydberg-calculator.readthedocs.io/en/latest/calculations_atom_pairstate.html#arc.calculations_atom_pairstate.PairStateInteractions.plotLevelDiagram))

Changes:
* 2.0.5. corrects offset due to hyperfine splitting in **Rubidium 87**. Now **transition frequencies** from **ground state to Rydberg states** should be within few MHz (relative to center of mass of hyperfine split lines) (f5319ccf64386b76aaf340bcf2db2d2861d4b709 and discussion in 20 ). This had no effect on Stark map or pair-state calculations, but it is important for experiments (finding transitions).
* since version 2.0.4. Python 3.7 is also supported (wheels are uploaded on PyPI) (19)