implement replication protocol between SM nodes

This is a big thing that we didn't get to in v1.

Between the time data is written to local storage and the time it is uploaded to cloud storage, there is only one copy of the data. So, if that node goes down that data is lost or at least unavailable.

Whilst implementing v1, however, there was general noise about improving CS's native replication capabilities. Need to investigate that more & think about whether or not it is also necessary to implement some replication functionality in SM as well. Might be redundant.