Deduplicate 'mapping_status' and mapping logic from functions called by map_harvest
Instead, make these functions only process what is specific to each tool type (ie. the layout of the metadata listing the files), and let map_harvest deal with what is in common (mapping_status + calling map_sha1_and_add_in_data)
This is what I meant in our chat on 2021-02-01 about deduplicating (and in the end, we don't need a generator to do it)
Migrated from D5079 (view on Phabricator)