Support Cursors & Batch Size

In certain conversations it came up that it would be great to get some kind of cursor on a N1QL response, very much like MongoDB and JDBC connectors provide it.

Of course this is different from using OFFSET and LIMIT, since let's say you do SELECT * FROM `travel-sample` and you want to analyze it all, getting back a cursor and then retrieving the data batch by batch with the given size. This would allow easy consumption of large amounts of data without overwhelming the clients (reading it all in one shot), better memory management and so forth.