[ ] Generate report of all duplicate images in image service (see admin function)
[ ] Classify them into 1. normal duplicates and 2. duplicates where different occurrence records (i.e. different scientific name or location) refer to the same image (see the koala photo which is assocated with lots of records)
[ ] For each class of duplicates, either delete the image and occurrence record (1) or remove the image reference from the occurrence record (2)
[ ] We might want to keep a separate Cassandra table of records that have been deleted, just in case we need to revert the change or the record comes back into the system from a fresh harvest (check with data team).