Create SystemConsumer and SystemProducer for HDFS

It would be nice to be able to read/write from HDFS, particularly for bootstrapping purposes. A few points:

• Per the discussion about leveldb this support should be separated into its own package and project (jar) for easy testing and severability.
• Similar to the Kafka RegexTopicGenerator, we can enumerate (recursively or not) the files in an HDFS directory during job startup.
• Connectivity with HCatalog would be interesting as well, but should be handled in a separate JIRA.