P3 processes

[rorlija1]

In-progress implementation/testing of P3 melting in KiD using code from this CloudMicrophysics PR.
And of shedding using this CloudMicrophysics branch.
And finally of collisions (I think my parameterizations are wrong though).
I've added the three rates to the tendency, but I'm using them individually because I have them in separate branches of CloudMicrophysics. Seems like they're all stable though!
Things to keep in mind for others:

• The testing configuration I've been using in KiD_driver.jl (dz, dt) seems to be the minimum dz / maximum dt to ensure stability... it's 36 vertical elements and a time step of 0.5 s, which seems very unstable, so I'm wondering if there isn't something at play in the code that's decreasing stability.
• p3_boundary_condition is a command line argument that allows for customization of the initial particle population
• I've added the "predicted variables" F_rim, F_liq, p_rim to aux, and given that we'll eventually want to compute those, plus thresholds, plus shape params, plus integration bounds each time step, maybe adding to aux a place for the struct that will contain those would be the way to go!
• I haven't added tests yet for this PR... sorry!