implement (optional) causal round-robin analysis in idq-batch

This will help us simulate "offline" results analogous to the "online" results. The associated logic should be relatively straightforward to implement, even within the existing set-up, but deserves some thought.

Chad mentioned this on the April 11, 2018 call as a low priority, so it's been assigned to him. In all likelihood, we'll probably implement this almost immediately after benchmarking the current batch infrastructure against previous results.