When you copy a page, you also copy its attachments, which leads to duplicated attachments on the Instance. Instead of duplicating the files, Confluence could handle on the back-end to identify if that file was being linked to different pages. That way, in case the file was deleted from one page, you can just delete the link - but keep the file. This would help to have smaller back-ups/export, in which sometimes a great part of its size is due to attachments - which can be duplicated.