Multiple decisions for a single manuscript version
I've noticed, looking at a recent Aperture dump, that some manuscripts have more than one decision in the database. This has mostly (but not always) occurred where at least one of the decisions was created by a user who is now deleted from the system (the user_id refers to a nonexistent user). This is dangerous, as there's no guarantee which decision will be displayed as the decision, and data may differ for each.
- Figure out why duplicate decisions were created, and fix.
- The fact that the user is missing in most of these cases may be a red herring, but if we permit users to be deleted, we should ensure that reviews etc remain functional, or else are deleted following due warning. I recommend that users shoudn't be deleted in normal circumstances; and if they must be deleted, then reviews etc remain in place but are labelled as anonymous.