Non reproducability (with different numbers of MPI processes) related to boundary conditions in implicit time stepping on unstructured

[kestonsmith-noaa]

Describe the bug
A clear and concise description of what the bug is.
When simulations use PDLIB_JACOBI_GAUSS_SEIDEL_BLOCK with boundary conditions rerunning WW3 with a different number of MPI processes will not reproduce the solution in a bit for bit manner.

To Reproduce
Steps to reproduce the behavior.
Run regtests/ww3_tp2.17 with different numbers of MPI processes and check against each other
Expected behavior
A clear and concise description of what you expected to happen.
Small differences in solution when run with different numbers of processes

Screenshots
If applicable, add screenshots to help explain your problem.

Additional context
Add any other context about the problem here.
The bug can be fixed with an additional communication of the state imeadiatly after the boundary conditions are applied within subroutine PDLIB_JACOBI_GAUSS_SEIDEL_BLOCK