It's odd to have the model (the set of equations) depend on the start and stop date of the simulation..
We need to do this now because the forcing data is part of the model, and the forcing data requires (via ClimaUtilities) the start and stop date.
We can get around this by moving the creation of the forcing objects to the simulation struct, and using "PrescribedERA5" struct "CoupledAtmosphere" etc instead, maybe