Change the Galacticus library to one based on EPS trees

[aphearin]

The current SED library comes from a Galacticus run on AlphaQ merger trees. This has numerous drawbacks that manifest in a variety of discreteness effects:

• Sparse sampling at the bright end since there are so few clusters in a 256 Mpc box.
• Sparse sampling at the faint end due to the finite resolution of the simulation
• Discreteness effects associated with the coarse timestep sampling of the simulation snapshots
• Galacticus color-color space has very sharp tracks in e.g., (g-r) vs. (r-i), as opposed to a smooth Gaussian-like spread
• Using N-body merger trees also comes with an increase in various outlier rates as differential equations become stiff due to discontinuities in the growth of simulated halos.

Since we are already ignoring the spatial positions in the Galacticus library, there is really no material advantage I see in using trees based on N-body simulations rather than EPS trees, which offer potential to substantially improve all of the above shortcomings:

• We can sample as finely as we want at both high- and low-mass ends, since there are no limits due to box size or resolution
• To improve redshift discreteness effects, we can use arbitrarily fine timesteps and/or use staggered timesteps of the galaxies in the library
• To fill out color-color-magnitude space, we can fill the library with models run with many different model parameters, sampling on an as-needed basis (as opposed to a simulation, where you only get the halos you get).
• Outlier rates should be reduced since EPS histories are monotonic and smooth, unlike N-body trees.

The one drawback that we have discussed with @abensonca is that we would not want to preclude the possibility of studying Galacticus-motivated assembly bias effects, which would be the case if the EPS trees only predicted halo mass growth. However, @abensonca is currently working on a formulation of the Galacticus model that responds to EPS trees with NFW concentration that are also under development.

CC @evevkovacs and @dkorytov