replace UniverseMachine ingredients with custom model

[aphearin]

Currently, the information our pipeline uses from UniverseMachine is through HOD-type sampling of the way the UniverseMachine galaxies occupy host halos in the MDPL2 catalog; this gives us M* and SFR-percentile for each galaxy.

The M*-dependence should be straightforward to mimic analytically, moving us away from the Monte Carlo sampling we're doing now. The two classes of empirical approaches I can think of would be:

1. use a CSMF-type analytic fitting function for the abundance of satellite galaxies as a function of host halo properties;
2. use analytic fitting functions for the occupancy of subhalos, then map M* onto the synthetic subhalos using a Behroozi/Moster-type function for the stellar-to-subhalo-mass relation.

Each of these approaches has a lot of of relevant development already in the current repo, so much of the effort here would be careful calibration of the fitting function parameters.

For the SFR-percentile, I think a simple power-law might provide sufficient flexibility; when I look at cluster satellites, their distribution of SFR-percentiles is just a moderately steep power-law heavily weighting the low-end near zero; when I look at 10^12 mass host halos that are rapidly accreting dark matter, their SFR-percentile distribution looks like a mild power law weighting the high-end near unity. Since there are already flexible functions for the double-Gaussians governing colors that are painted on top of these SFR-percentiles, then this power-law relation for the SFR-percentile distribution may be flexible enough to control the two-point function and cluster-quenching.