Add a null sink option for the consumer
When evaluating throughput of the flink pipeline it would be nice to remove elasticsearch from the equation. Our production clusters are able to sustain much higher throughput than the testing clusters that are available, this should make it easier to verify backfill capacity of the flink pipeline itself without having to worry about elasticsearch.
Change-Id: Ie6e15ebc2d24cbe75fac579eee0fdcf72c6a9efc