It would be useful if there were a way to set default quotas that would apply to all future users.
I'm sure there are other situations where it would be useful, but the one I've encountered involves Singularity (Apptainer) where files are created and owned by subuids. The subuids are nominally chosen from a minimum pool of 65536 for each user and the files are created using different subuids as users use Singularity so there isn't a good way to predict when or which subuids will create files. A way to set a default limit for a particular filesystem would solve this.
If there is already a way to do this, then it would help to add some discussion in the man page.
Anonymous
Well, the decision to set quotas was always on userspace - e.g. your user creation script to call setquota(8) or something like that. Now what you ask for would either have to happen inside Singularity (it certainly needs to track somehow used/unused subuids and it could set quota when allocating new subuid) or it could happen inside the kernel when we notice some UID is doing allocation for the first time and we are creating structure in our space tracking. Quota tools (whose tracker you are using ;) does not have much say in this.
XFS actually has a feature which you could presumably use - default quota - UIDs not having quota limits set can be limited by a configured defaults - see xfs_quota(8) manpage for details. For any other filesystem it would mean a change in the on-disk format of quota tracking structures -> not so easy to implement (but I can imagine implementing something like XFS default quotas for e.g. ext4 if there's sufficient demand). Also the kernel solution has rather limited flexibility - I can imagine people soon wanting different quotas for different subuids (or users), wanting control over usage of all subuids of a user etc. - and kernel cannot easily provide this.
So IMO the most flexible solution for the future would be to implement quota support in Singularity but of course I may be biased by not being Singularity developer ;)