Shared NFS means yo u have the latest VM/CT state instant, with ZFS replication the VM disk state/content may be 15 minutes old (or whatever your smallest replication interval is).
Also, the whole disk is replicated on all nodes where you have setup a job, if you have a single NFS the VM/CT disk is stored only once (depending on your NFS, this could also not be ideal if it does not provides some sort of reliability/failure correction measurements).
So to give a better answer we'd need to know what your NFS setup would look like.