More QC - LL per iteration, histogram of fitnesses

[darachm]

Having immediately available QC helps users iterate and make sure they got the right and stable fitting configuration without having to spin up additional analysis software.

Here's two to do:

1. This is copied from: Feature request - log-likelihood output file FangfeiLi05/PyFitSeq#10
   It is useful to see the trajectory of changes in the log-likelihood to analyze the kinetics of the optimization.
   Presently, I think that's just output to stdout.
   Let's add an argument, if set, will write LL to a file (set by the argument of course!!!)
2. This is copied from: Feature request - QC histogram of fitnesses FangfeiLi05/PyFitSeq#9
   It's often diagnostic to see how the fitnesses are distributed. Sometimes they can look funny,
   indicating parameters are converging strangely. A diagnostic histogram of fitnesses would be a handy output!