I can't find a JIRA for this, though there are some related to the existing
MLlib implementation (https://issues.apache.org/jira/browse/SPARK-10802 and
https://issues.apache.org/jira/browse/SPARK-11968) - would be good to port
it over, and in the process also speed it up as per SPARK-11968, and
I have experimented very low performance with the ALSModel.transform method
when feeding it with even a small cartesian product of user x items.
The former mllib implementation has a recommendForAll method to return topn
items per users in an efficient way (using the blockify method to distribute
