Add ocean model conservation checks

[cbegeman]

This PR adds only the conservation checks done in MPAS-Ocean, with commented-out placeholder code for the non-boussinesq mass conservation check for Omega.

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]

@xylar Can you provide really broad feedback on the structure of this PR, particularly

• Treating conservation checks as an instance of a property check
• verify_properties as argument to add_output_file
• verify_properties as a method of step which is overridden by OceanModelStep
• whether to rename verify to check everywhere

[xylar]

@cbegeman, this looks great!

• I like the idea of a more general property check, of which the 3 conservation checks are a subset
• I'm fine with verify_properties as argument to add_output_file, and as a list of string names of properties to check.
• I'm good with verify_properties being a method of Step, overridden by OceanModelStep.
• I think I do somewhat prefer check to verify so if you want to change that, I would be in favor. But I don't feel strongly about it.

Let me know when you're ready for a more detailed review.

[cbegeman]

@xylar Great. Thanks for taking a look! I'll try to finish this up today so we can get it in before the break

[cbegeman]

Testing

I have run the mass conservation check for the barotropic_gyre case for both MPAS-Ocean and Omega on chrys with intel, openmpi

[cbegeman]

@xylar Let me know if you would like me to exclude 38a167d

[cbegeman]

@xylar Ok. It think this is now ready. I realized that it made sense to support conservation checks for all omega-ported tests.

[xylar]

@xylar Let me know if you would like me to exclude 38a167d

Nope, I'm fine with sneaking those fixes in here.

[xylar]

My main non-cosmetic comment is about what layerThickness means coming from Omega. This actually may be something we need to address in a separate PR. It may be that we need layerThickness <---> GeometricLayerThickness and we need to have a way to compute Omega's LayerThickness from GeometricLayerThickness as part of computing initial conditions. But I want us to have a plan for that because this PR will somewhat exacerbate that problem.

And, as I say below, I think we should strongly consider renaming LayerThickenss to PseudoThickness in Omega.

[cbegeman]

we need to have a way to compute Omega's LayerThickness from GeometricLayerThickness as part of computing initial conditions. But I want us to have a plan for that because this PR will somewhat exacerbate that problem.

@xylar Does this PR exacerbate that problem if we're calling separate functions for MPAS-O and Omega, e.g. compute_total_mass_nonboussinesq? I think you made a good point that we may want to use PseudoThickness rather than geometric layerThickness for Omega. But are you thinking that we would really want to convert to layerThickness for Omega for the tracer conservation checks?

[xylar]

@xylar Does this PR exacerbate that problem if we're calling separate functions for MPAS-O and Omega, e.g. compute_total_mass_nonboussinesq? I think you made a good point that we may want to use PseudoThickness rather than geometric layerThickness for Omega. But are you thinking that we would really want to convert to layerThickness for Omega for the tracer conservation checks?

I think this mainly exacerbates the problem if we don't account for layerThickness = PseudoThickness. As long as we account for that, we're good. I'll need to fix things in #440 to make sure we don't use layerThickness there but rather PseudoThickness.

[cbegeman]

@xylar Do you want to take a look at the last 2 commits and let me know if I've made the changes you had in mind?

[xylar]

@cbegeman, great! I think your solution to adding a helper function that takes care of all cases simultaneously quite clever!

Looks great to me!

[cbegeman]

@xylar Thanks for reviewing and getting started on some related draft PRs