Convert SPARK example from Markdown to Jupyter notebook

[chmreid]

The SPARK example in the repo is currently in Markdown format and is written as a walkthrough. We should convert it to a Jupyter notebook, and ideally find some way of testing the notebook.

(Side note: minikube allows kubernetes code to be tested on Travis using a single node, perhaps there is a minispark or something similar to allow us to test Spark code on a single node.)