Comment (by Roland Haas):

Zach may be ignoring the level 1 warning. For this usage case, he can possibly do so (though I am not sure since the results of CoordBase_GetBoundarySizesAndTypes will differ on different MPI ranks since it returns the ghost size for patch boundaries that are not physical boundaries but instead inter-processor boundaries.
@Erik Schnetter @Zach Etienne you may need to clarify the suggested code and observed behaviour.

--
Ticket URL: https://bitbucket.org/einsteintoolkit/tickets/issues/2288/leans-boundary-condition-setting-for-bssn