Following on from #39, in which package-level stats were autogenerated, we need to also generate these stats at the repository level. This will necessarily have duplicates that get unioned. These stats should include:
[ ] Number of unique parts
[ ] Classes of parts
[ ] Number of plasmids used
[ ] Number of part/backbone combinations (i.e., wells required)
Following on from #39, in which package-level stats were autogenerated, we need to also generate these stats at the repository level. This will necessarily have duplicates that get unioned. These stats should include: