Fix small conservation error when using fct thickness and tracer advection

[trhille]

Add a halo update on layerThicknessEdgeFlux to fix a small tracer conservation error when using fct thickness and tracer advection. That error was documented in this comment in PR #70 : #70 (comment).

[trhille]

Testing

Using the grounded slab advection test with prescribed velocity that I used extensively in #70. Calculating tracer volume using ncap2 -s 'tracer_sum = (passiveTracer2d * thickness * areaCell).sum($nCells)' output_all_timesteps.nc tracer_sum.nc. Before this change, we got a drift in tracer volume:

Using this branch, we get full conservation:

[matthewhoffman]

@trhille , good sleuthing. This looks straightforward. It's too bad we have to pass domain around just to do a halo update, but solutions to address that are well beyond the scope of this PR and the solution you have here is the obvious one.

[trhille]

All full_integration tests passed validation and baseline comparison.

[trhille]

All fct_integration tests pass validation. landice_humboldt_mesh-3km_restart_test_velo-fo_advec-fct_calving-von_mises_stress_damage-threshold_faceMelting fails baseline comparison, presumably because now damage is treated properly.