Feature request thread for supporting new ways of controlling POD execution

[tsjackson-noaa]

This is a thread for requests for the package to support different ways to control POD execution: protocols to allow the framework to start the PODs, and to transfer control back to the framework when the PODs have finished running.

Currently this is done by starting each POD in its own subprocess on the same machine the framework is running on, but one can envision interfacing with slurm or other job schedulers, as well as more fully featured workflow pipelines, either on a local cluster or in the cloud.

The MDTF package is developed as a community effort, and we need community input to prioritize how we should allocate developer resources. Community contributions are even better -- if you've adapted the framework to run PODs in a new way, please submit a pull request!