Plan for scaling up workflow for larger applications

[cahhansen]

Things to consider:

• Storage of NWM retrospective data (~7000 reaches correspond with USGS gages (needed to compare model performance)). This is not terribly time-consuming to process (only needs to be processed once), but it's important to think of what is the most intelligent way to store/manage the time series for all reaches. How big is an Rdata file with all 7000 reaches?

• For scaling the use of these functions to large areas, we may want to provide versions of the R functions like NWMFlowStats in Python to work nicely (compute this on the fly?) with visualizations such as the Low Flows Tethys Application.