Aggregator interface

[willscott]

In thinking about extensions to storage deals, there is interest in having a common aggregation interface, such that there are more options for where small deltas of data can be submitted for storage into filecoin.

One variant of this would be that clients can send sub-pieces of data, knowing their sub-commp for that slice of data, and the SP gets the storage collateral once they make a deal where they can show there's a merkle inclusion of that sub-commp.

Another extension to this is cases like compute over data, where for latency reasons there would be a significant value in external entities being able to know which sub-pieces have been received and are awaiting deal inclusion (and which clients have failed to send the data they were supposed to and should be re-scheduled)