ZFS can't delete files when over quota

[grarpamp]

NAME                     USED  AVAIL  REFER  MOUNTPOINT
tank/distfiles          10.0G      0    25K  /tank/distfiles
NAME            PROPERTY               VALUE
tank/distfiles  quota                  10G

What is your free in 'zpool list tank'?
Because of CoW for metadata, I think it makes sense that
your tank/distfiles won't rm within your used up 10G limit there.
What if you raise the distfiles limit to use more of tank?
What if you rollback distfiles?
And consider using per user/group quotas so root can still
rm in tank/distfiles [1].

[1] Sure, ZFS arguably should make internal reserved
space to allow rm (the effective metadata). Question is,
how much, what dynamic metrics for that? Maybe ZFS
feel admin still has some responsibility. But you really
don't want to take a user ticket because they can't
manage their own rm as under other file systems.
What is the state of this in current Solaris?