Fixes to landlocked cells when culling global meshes

[xylar]

Since critical land blockages can cause new cells to be landlocked, we need to remove landlocked cells only after we have added land blockages to the land mask.

This merge also fixes a small issue where the latitude_threshold parameter for widening critical passages is now supplied from the appropriate config option, whereas it had previously been hard coded by mistake.

Checklist

• [x] Document (in a comment titled Testing in this PR) any testing that was used to verify the changes

[xylar]

Testing

I ran the QUwISC240 mesh test case with this branch. Previously, a single cell on the coast of the Eastern Pacific was marked as landlocked only in the second pass (which is meant to "detect landlocked" cells at ice-shelf calving fronts). This meant that it was marked as an ice-shelf cell (landIceMask = 1) instead of a land cell (culled from the mesh).

With this branch, the cell is now correctly identified as land and culled from the mesh.

[xylar]

I verified that there are no such cells in the IcoswISC30E3r5, RRSwISC6to18E3r4 or SOwISC12to30E3r2. 😌

[xylar]

Thanks for taking a look, @cbegeman.