Use microbenchmark to benchmark render pass

[steve-the-edwards]

trafficstars

We can use a similar WorkflowNode test as for the jmh benchmarks, but using microbenchmark will allow us to easily run on a mobile H/W device rather than on the JVM on the development machine. Obviously Workflow performance matters beyond just mobile, but that is the most significant current context.

[steve-the-edwards]

Removing this from baseline milestone as optimization internal to render pass is not needed at this time. How many times/how we render pass is more crucial.