Update code to use Dataflow's new support for custom sources

[deflaux]

We manually create data shards right now via --references (or --allReferences) and --basesPerShard from ShardOptions.

Updating to custom sources will allow the data shards to be not only createed but also re-sharded dynamically.

https://cloud.google.com/dataflow/java-sdk/JavaDoc/com/google/cloud/dataflow/sdk/io/Source

https://github.com/GoogleCloudPlatform/DataflowJavaSDK/blob/master/sdk/src/main/java/com/google/cloud/dataflow/sdk/io/Source.java

[pgrosu]

That would be awesome! Yeah, I mentioned something along these lines a while ago here, and very excited to see it:

googlegenomics/spark-examples#49 (comment)

Maybe the size of the requested region can be processed by a function to dynamically return and define the --basesPerShard.

Thanks and very excited to see the results!
~p