interpolation index calculation in update_UVbackground_rates should be smarter

[brittonsmith]

We currently use a brute force loop to calculate the index for redshift interpolation in update_UVbrackground_rates. It's not a huge deal because the tables are only of length ~20, but they are evenly spaced in log(1+z). We should calculate this arithmetically or at minimum do bisection.

[mabruzzo]

FYI: A length of ~20 may be near the limit where linear search is faster/comparable to binary search/bisection

[brittonsmith]

That's a good point. Maybe we could just switch it to arithmetic assuming even bin spacing.