Discussion related to implementing CCPP-framework provided process- and time-splitting via SDF

[grantfirl]

The evolution of the state within the physics loop should be a decision for the host model, and not of any parameterization. This is complicated by time-split vs. process-split schemes.

Process-split physics processes are schemes that share a common input state, whereas
Time-split processes use the state provided by the previous physics process.

Questions/Obstacles:
Is this a responsibility we want offloaded to the ccpp-framework?