hamlet does not process mutex ruptures correctly

For JPN, where we expect ruptures M>9 in the source model, the model Mmax=8.4.
Using a job with only one mutex source containing several nonparametric ruptures, I found that no sources are returned by get_rlz_source. I tested what would happen if I add the following:
if len(srcs)==0:
grp = csm.src_groups[rlz]
srcs = [s for s in grp]
This works for returning sources, but gives errors due to having 0 ruptures (based on _source_info). Changing s.num_ruptures to s.count_ruptures() in lines 88 and 104 of the source_reader, hamlet will finish and create the expected plots, noting that the mutex rupture weights have not been applied and their probabilities have not been scaled to 50 yr investigation time (i.e., the rates are way too high relative to the observations).

However, keeping these changes, if I run the full ssmLT, then the original problem persists (model Mmax is again 8.4)

I'll send a test job separately