Initial Spark Support (Port of Scalding)

[tixxit]

This isn't merge ready yet, but since I know a few other people are interested (eg @non), I figured I'd make the PR now so I can get some more eyes on it.

A few thing worth mentioning:

• Spark doesn't really have support for something like Execution, so we just use its strict Unit-happy API everywhere.
• We only write to files when explicitly asked and use .cache() for memoizing the trees in-between steps.
• Basically just implements expandTimes and expandInMemory, but not prune or validate (not hard, just haven't done it).

[tixxit]

You can run the example with something like: SPARK_HOME=/path/to/spark ./iris-spark

[tixxit]

We have to do this because Algebrid's MutableBackedMap isn't Serializable!

[tixxit]

Soon we can get rid of this: twitter/algebird#424

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}