Add support for reanalysis-driven steady-forcing single column runs and global time-varying configurations

[Julians42]

PR #3758 adds support for single-column PROSPCT driven by time-varying reanalysis forcing. Eventually we would like to also support the following two features:

Steady Forcing Driven by Reanalysis for Single Column Models

While this was drafted by @Julians42 in #3519, review suggested that additional work is needed to refactor the code to comply with current ClimaAtmos standards. With #3758 possibly the refactoring could be folded into this framework. Steady forcing has the disadvantage of not resolving the diurnal cycle but would be more data efficient and is simpler than the time-varying reanalysis forcing to implement and understand physically.

Globally driven Reanalysis framework

@oalcabes is interested in experimenting with observationally driven global runs. Driving global runs with reanalysis might fit well within the calibration framework as it might allow for shorter runs or calibration to different metrics. It is unclear how much effort beyond the time-varying single column model this would require. It is not a priority at the moment and should be discussed with @tapios.

Support for truncating domains

At one point we wanted a feature whereby you could run a model with the stretch grid from a larger model but only up to a certain height - e.g., run the stretch grid from a 70km model but only up to 40km, mostly for computational efficiency. I drafted this here but gave up on merging it a while ago.

[imreddyTeja]

@Julians42 Is this closed by PR #3758 ?

[Julians42]

@imreddyTeja These are plans for future model development we are considering that are similar to #3758 but weren't addressed by the PR.