Substitute `.splups` files for `.scups` files in functions for older versions of the database

[jwreep]

#17 had a note that the files will need to be substituted in any function that makes use of .scups datafiles. The tests would need to be suitably modified as well.

[wtbarnes]

I wonder if we should have some higher-level object that exposes the information provided by these files. I'd like to avoid having to deal with if splups exists logic all over the codebase as I imagine it would become very messy very quickly.

I'm just thinking aloud here, but we could have some sort of additional attribute _upsilon_data attribute that prioritizes one versus the other. We may need to abstract away how the spline fitting is done as well. I'm not exactly sure how similar these two darasets are.

[jwreep]

That's probably the way to go. Considering #11 and #235, this might be a good time to think of how to generalize all the data input. If one wanted to use atomdb's data to calculate level populations, for example, what modification would the code need?

[wtbarnes]

I've been thinking about this a bit more after dealing with #426. Having some abstraction for "processes" (like electron and proton collisions) would make changing the underlying data source much easier as well as limiting what levels these processes apply at.

If/when implemented, the number of levels in the splups files also needs to be factored into the number of levels present in the atomic model which is now computed on the n_levels property.