findBatch doesn't work with RX API

[diabolicallabs]

The findBatch operation calls the result handler multiple times with each document in a batch. When finished, it sends a null object. This doesn't work with the RX API as it generates code that returns a Single since the result handler only mentions one document. So all we ever get is the first document instead of an observable stream of documents.

Perhaps a batch read stream should be created instead. Actually a batch write stream would be nice too. Such an implementation would diverge from the current service proxy api but would be very useful in dealing with large datasets with RX and in general.