Add Spark Streaming with Kinesis support

[alexanderdean]

See #43, #139

The basic idea is that you point a new webhook at a Snowplow Scala Stream Collector. Behind the Snowplow collector, we have a Spark Streaming job running Schema Guru. We would shard based on schema so that only one worker is operating on each schema. We fetch the most recent schema from Iglu, merge the output of our new 5 minutes worth of derivation, and then push the updated schema back to Iglu as a patch.

/cc @chuwy @fblundun, @a1nayak