wrong from_sigma8() when a_stop != 1

[adrianbayer]

To compute sigma8 we need growth at a = 1 in the varlin function, but pmwd only integrates the growth for a <= a_stop. So the sigma8 conversion by from_sigma8() is wrong when running with a_stop < 1. Quick fix is to use a conf with a_stop = 1 to compute A_s, and then use the conf specified by the user after -- or to just always enable computing the growth up to a = 1 regardless of a_stop.

For example, the below plot shows growth(a) for a conf with a_stop=0.7 and a conf with a_stop=1. If a_stop=0.7 is used then growth(a=1) is incorrect and thus from_sigma8() will give an incorrect A_s and final power spectrum.