Overhaul user tagging for resources

Tagging LoRAs has several big issues. The big one is that the user is never given the count when typing tags and is suggested any tag with any prior use. This means we have spread resources across many redundant tags and defeated the purpose of tagging (besides making them search keywords). It does no good to have “comic” “comics” “comic book”, “comicbooks” etc. all be their tag. I propose the following
1: Show users the current uses of tags when suggesting them in auto complete (and possibly while tagging)
2: Have a way to list model tags by use count because, as far as I’m aware, it doesn’t yet exist and thus impossible to know what are useful tags to include
3: Add a system for users to propose merging tags to clean up the list of all the slight variations we’ve already got.