For the record, the follow-up question is covered in a new thread:
https://root-forum.cern.ch/t/discrepancy-in-2d-histfactory-fit-results-between-root-versions-6-28-06-vs-nightlies-with-beeston-barlow-method-activated
In general, it’s better to open new threads for follow-up questions once the original question is solved, because then the post is not on our radar anymore. So it’s good that you did this!