Runtime modeling

[ian-bertolacci]

I have a framework for modeling footprint with pretty good accuracy (for the data it's trained on), but do not yet have one for runtime.
Currently, accuracy is not a concern, as our current data set is limited to changes in x and y (and some z), but not time-steps or processes.

Deliverables:

• Several methods of modeling parflow runtime from our existing data.
• At least one model that will be used in the current tools default implementation

[ian-bertolacci]

Should note, this will be done using existing data, so no new runs will be needed for this initial testing.