JanusGraph OLAP on vanilla Spark cluster

TinkerPop has a bin/hadoop/init-tp-spark.sh to distribute library jars across the nodes of a vanilla Spark cluster. This script does not work for JanusGraph and a similar script adapted to JanusGraph + documentation would help certain users.

This issue was created as as result of [this thread](https://groups.google.com/forum/#!topic/janusgraph-users/ICREKjSUw8M) on the JanusGraph users forum.

Start by comparing TinkerPop's bin/hadoop/init-tp-spark.sh with how JanusGraph libraries are distributed across a vanilla Spark cluster. Review the linked JanusGraph users forum thread for the compatibility requirements. Done means a JanusGraph-adapted script works for the described cluster setup and accompanying documentation explains its use.

Written by the indexing model from the issue text.