Regional boundary data lacks pressure on non-scalar velocity points

[SamuelTrahanNOAA]

Description

The fv_regional_bc.F90 requires surface pressure on all velocity points for wind remapping. Velocity boundary points go beyond scalar boundary points, and the surface pressure is only available on scalar boundary points. Surface pressure must extend beyond velocity points, to i-1 and j-1 locations relative to velocity points. That means the surface pressure boundary regions must be larger than they are now, by 2. We also need the surface pressure "boundary region" to include one point within the domain. I don't know where this issue originates, but I know it is a problem in the regression test input data.

The code should only be reading surface pressures in regions where it already has that data. Reading outside those regions is a bug; the extent of the boundary is not at fault.

Edit Sep 27: Surface pressure is needed beyond velocity points.
Edit Oct 3: Needing surface pressure beyond those points was a bug.

To Reproduce:

Read the files, OR initialize ps_reg to real_snan to fv_regional_bc.F90. Eventually, after fixing a series of other bugs, you'll encounter invalid surface pressure data on velocity points.

Additional context

FIXME: Link to other issues and PRs

NOAA-GFDL/GFDL_atmos_cubed_sphere#218
NOAA-GFDL/GFDL_atmos_cubed_sphere#219
NOAA-GFDL/GFDL_atmos_cubed_sphere#220