Enhance benchRun to not only take multiple ops, but also specify number of threads per op

I want to run two workloads in parallel, one doing reads and one doing writes, with a different number of threads doing each op. The best I can do is start two workloads via benchStart, sequentially, and then wait X seconds, and then call benchFinish sequentially. This hack is not very scientific, as launching a large number of threads can take a while and one workload is starting before the other.