Not sure about this, but an (educated) guess:
Replication locally on the machine can easily be implemented by a multi-threading approach, but the question is if this would give you much benefit as the data-transfer to/from the HD's is very much a sequential process. Of course you can do multiple HD's in parallel, but synchronization between the parallel processes would be very important and complicate matters a lot.
When you throw a LAN connection into the mix matters are again different.
The SSH connection is a single threaded thing. You can obviously run multiple connections in parallel but again the nature of a LAN connection is highly sequential (and LAN speed usually the most limiting factor) , so it is debatable if this would gain much performance and whether the effort is worth the gain.
And of course... Is the other side able to handle this in the same way ?
If the receiving NAS has less cores (worse case: just 1) than the sender you can easily saddle it with an enormous amount of task-switching-overhead.
And if the physical disk-structures between the NAS-boxes don't match there is also an (non trivial) amount of data re-organisation that needs to be done on the fly.
Again: Would it be worth the complexity and effort ?