improve current [groom] processing patterns

[cchriste]

Current [groom] processing patterns create i/o bottlenecks. The various steps are carried out separately. Each step reads one file at a time, runs a command on that file, and writes the result. However, for most of these processes a file could be read and all steps executed while it's in memory, and the result written at the end.

Furthermore, the process could be carried out in parallel (especially if significant processing is required for any step and it the step has no function-level parallelism).

One reason to execute the steps independently is that there is sometimes usefulness in looking at mid-process results. We should consider a way to accomplish both application-level parallelism as well as how one might consider these mid-process results.

This is not urgent, but should be on our radar.

[sheryjoe]

Related to #865

[cchriste]

This is essentially how/why we want to get rid of things like GroomUtils, for which we may already have independent issues.

[jadie1]

Done with PR #1181