Add reduce option on accumulo-spark ImportRDDOfElements

[james010101101]

Currently the importRDDOfElements operations just partition's the elements according to the accumulo table's split points and writes the result to HDFS to be imported, We should add an option so that the RDD can be aggregated before partitioning.

[p013570]

@james010101101 is this still required? Have you done any work on it?

[gaffer01]

As noted above, this should be optional. The default option should be to perform the aggregation (as long as there are some groups with aggregation enabled).