There’s a well known approach in software involving a severe avoidance of ever using uman-provided names as the canonical identifiers of any resource (e.g. a group). We do this for lots of things or possibly even all things inside Ardour. Essentially, the software doesn’t care what you call it, because we have our own identifier (*).
This, however, leaves open the separate question of whether or not we should hold your hand and stop you from doing things that might make your life more difficult (like naming two groups the same). At this time, for groups, we do not do that.
(*) one notable exception is stuff that affects I/O ports, like track names, because if you use the JACK backend, the names are required to be unique