If you have a number of different repositories (which is very common if you're using Git or another DVCS where often you'll have one repository per project), and a lot of users, you have to do a LOT of User Mapping entries to support that type of use case.

So if I have N users, and M repositories, I need to have N*M manually added entries in the User Mapping table in order to support all my different mapping points.

It would be much easier if I could just have N entries (with "all repositories" set), and maybe a few overrides or extra ones for repositories where that doesn't match properly (though that's really just an arbitrary option to be honest).

If you're not using Crowd this is actually much easier, because you can set a default mapping for each user that spans repositories. If you ARE using Crowd, though, your only option is the N*M entries in the User Mapping interface.