storage/client: Only wait for F+1 successful writes

[tjanez]

After #1792 is merged, the storage client will utilize all the nodes from the storage committee.

However, when writing to nodes, it will wait for all the nodes to finish. We should implement only waiting for F+1 nodes, where F is a consensus parameter that needs to be checked by BFT roothash (as otherwise some compute nodes could commit to less than F+1 nodes).