Hey, is it reasonable to have a dedicated repository for ztokens, instead of introducing a column for it in every document? The dedicated model would hold: - resource type - resource id - the ztoken This model would be changed in the same unit of work as the related document(s) for atomicity. I can see this simplifying things a lot for us, but I just want to pop this question to understand why it is generally recommended to store it alongside the document. Am I missing something?