I have some code in the jointcal notebooks to read in the "chi2 contributions" files: we need something similar in a jointcal module to make it easier to work with the many new such files that will be available. Besides just reading the files as astropy tables, we should have a simple data structure to make it easier to get at values of interest and to collate all of the different outputs (e.g. linking measured and reference sources, linking objects from one step to the next). In addition values like the total chi2/ndof (which should be directly computable from the output files: we can double check that against the values in the log) should be computed immediately for each file we read in, and the outliers that were flagged in the "inner" fit loops should be extracted so they're easily accessible.
I think it might be good to have a plotting API live on top of this, to organize all the different types of plots and plot options.