Sparse Coagulation Approximation

[Gorkowski]

I would like a reduced resolution (fewer radii points) for the calculation of the coagulation rate so it doesn't cost as much time.

My current thinking is to replace the current coagulation rate (gain and loss) with fewer radii and then upscale back the original distribution indexes.

Originally I was thinking of doing this in coagulation_rate.py, but after testing some options, it might make more sense in rates.py

And any obvious errors this may introduce in the integration must remain small and random

[ngam]

Two quick thoughts:

1. Usually we don't need to go lower than 0.1 nm increments for really high-quality results
2. We may be able to cheaply do this via:
   a. Some lookup table (or pertained model so to speak)
   b. Using simple splines, e.g., for radii between 10 and 100 nm, calculate sparse coag between 5 and 200 nm, then use splines to get the ones between 10 and 100 nm at whichever resolution you desire ---

(something important to keep in mind is interpolating along varying size when the other size is fixed, otherwise, one could use interp2d)