non-BFB results for ocean/drying_slope/1km/single_layer/ramp when using more than 9 procs

[gcapodag]

The results for the normalVelocity are not BFB when comparing a serial run with a parallel run using 10 or more processes (I tried 10,11,12,13). Using 2,3,...,9 procs, the results for the normalVelocity are instead BFB.

[xylar]

@gcapodag, I think this should also be an E3SM issue, since it is presumably a problem with MPAS-Ocean standalone, not with compass?

[xylar]

But there's no problem with having issues in both places.