When different versions of the deblender and/or measurement catalogs are executed using the same mergeDet catalog, meaning the same parent blends with the same footprints, the combination of (parentID, peakID) for each source is unique, meaning we can exactly match all sources between the catalogs.

This will be more accurate than matching on centroids, and will also allow us to have a more accurate plot on how centroids are different for the same source in different deblending/measurement catalogs.