Modified ocean/mesh/remap_topography to allow smoothing

[bmooremaley]

The ocean/mesh/remap_topography step has been modified to remap combined topography from a cubed sphere source grid to MPAS meshes, instead of a lat-lon source grid. The weights file generation has been switched to mbtempest so that smoothing can be defined in the target MPAS mesh SCRIP file. Smoothing parameters have been added to ocean/mesh/remap_topography.cfg. This new workflow also uses a pre-build source SCRIP file stored in the bathymetry database, in addition to the combined topography file.

Checklist

• User's Guide has been updated
• Developer's Guide has been updated
• Documentation has been built locally and changes look as expected
• Document (in a comment titled Testing in this PR) any testing that was used to verify the changes

[xylar]

@bmooremaley, as we discussed, we'll need some way to support a different topography dataset for lower res regression testing. I pulled out the config options into their own config file:
#864
If we want to continue to support that 1/20 degree dataset, we would need a way to still use ESMF for that resolution. An alternative (if mbtempest if fast enough) is to use a lower resolution cubed-sphere dataset as we discussed and stick with the modified workflow you have here. I would lean slightly toward the latter so that the regression testing is using a more similar workflow to the high resolution runs.

[xylar]

I'm moving 5de71c4 to #879. This branch should be rebased once that PR gets merged.

[xylar]

@bmooremaley, some preliminary comments and suggested changes while you work on the documentation.

[xylar]

@bmooremaley, a few more suggested edits. I'm extending this code in #829 to combine this topography with a dataset from MALI. That process is qualitatively similar to your modified remap_topography so it was a useful chance to look through your code in more detail.

[xylar]

I think the only remaining steps are:

• run all the global ocean tests from the pr suite but without cached meshes and initial conditions
• update the cached mesh and initial condition files
• make sure the pr suite succeeds

[xylar]

Testing

I tested this on Chrysalis with Intel and OpenMPI. I ran the Icos240, IcoswISC240, Icos and IcoswISC meshes through the mesh, init and performance_test test cases without caching. Then, I cached mesh and init steps. Then, I successfully ran the pr suite (which uses the new cached files). I looked at the relevant topography fields from the initial_state step for each mesh and everything looked great to me.

[xylar]

@bmooremaley, thanks for very hard work on this! I took care of the last few finishing touches so I think it's ready to merge!