Decide which category is chosen from multiple duplicates with the same datasource when merging

[janehmueller]

Currently when a Subject has two duplicates from the same datasource and both these duplicates have a category, while no Subjects with a "better" datasource have a category, the values of the category are sorted and the first one is chosen. It would be best to sort the duplicates of the same datasource by their duplicate score first before choosing the correct value for the category.