Enhance shared convergence steps

[cbegeman]

This PR makes it possible for all convergence tests to leverage shared convergence steps and implements those shared convergence steps in all convergence tests.

The general approach is that spherical test cases will invoke SphericalConvergenceForward, which descends from ConvergenceForward, and ConvergenceAnalysis and that planar test cases will invoke ConvergenceForward and ConvergenceAnalysis.

The config options are moved into convergence and convergence_forward and only the default resolutions remain in spherical_convergence_forward.

Checklist

• User's Guide has been updated
• Developer's Guide has been updated
• API documentation in the Developer's Guide (api.md) has any new or modified class, method and/or functions listed
• Documentation has been built locally and changes look as expected
• Testing comment in the PR documents testing used to verify the changes
• New tests have been added to a test suite

[cbegeman]

Testing

I have run the following convergence tests on chrys with intel, openmpi:

• ocean/spherical/icos/cosine_bell
• ocean/planar/inertial_gravity_wave
• ocean/planar/manufactured_solution

I reran the convergence tests on perlmutter with gnu, mpich.

[cbegeman]

@xylar Can you take a look at this when you have a chance and let me know if you are on board with these changes before I write up the documentation?

[xylar]

This is brilliant! Definitely a good demonstration of the shared-framework philosophy behind Polaris.

I have a few finicky suggestions but the approach looks excellent to me.

[cbegeman]

@xylar Thanks for the detailed review! I'm on board with all the changes. I'll make them and retest this week.

[cbegeman]

@xylar This is ready for your (re)review. I also added a convergence suite which is convenient for testing changes to the convergence steps.

[xylar]

I ran the convergence suite and compared with a baseline from main on Chrysalis. The cosine bell tests are BFB with their counterparts from the cosine_bell_cached_init suite from main. Baseline comparison fails for the planar convergence tests because the number of time slices in the output has changed (from 31 to 2). Since I didn't see an easy way to work around this and the convergence rates look as expected, I didn't pursue this further.

[xylar]

@cbegeman, if you want to either clarify or remove the print statement I flagged above, I think that's the last thing needed before this can be merged.

[cbegeman]

@xylar Thank you for your review! I removed the print statement and will merge once the checks are finished.