updated the max redshift in a conditioned distribution
this may be dangerous because it is not clear how well the approximation made within ParetoFlatComponentMassConditionedOnRedshift.z2min_bh_mass() will hold at high redshift. However, we're hoping it should be good enough.
For posterity, the assumption is that
m1/Msun >= min(max_bh_mass, max(50*z, min_bh_mass))