Create a set of Carrot tests for FlowBasedGATK

[jamesemery]

We already have a battery of carrot tests that we would like to run to evaluate the HaplotypeCaller. However those tests do not currently cover any of the new use cases in the FlowBasedGATK code. The ask here is that we should create and design a set of longer running, multi-sample evaluation code for the FlowBasedGATK on a reasonable set of non-sensitive input bams that we can call and then subsequently run evaluation metrics on so we can have a better evaluation of how future improvements impact the FlowBasedCode.