Fix MPI BdIntegral on deformed spherical shells

[gthyagi]

Summary

• treat empty local boundary strata the same as missing boundary labels in UW_DMPlexComputeBdIntegral
• add an MPI regression test covering BdIntegral on a deformed spherical shell after _deform_mesh()
• keep existing parallel annulus boundary-integral coverage passing

Verification

• mpirun -np 2 /Users/tgol0006/.pixi/bin/pixi run -e amr-dev pytest tests/parallel/test_0765_internal_boundary_integral_mpi.py -q --with-mpi
• /Users/tgol0006/.pixi/bin/pixi run -e amr-dev pytest tests/test_0502_boundary_integrals.py -q -k spherical

[lmoresi]

Fix is correct — checks local stratum size before entering the Allreduce path, handling the general case of empty local boundary strata under MPI partitioning. Test covers the deformed-shell scenario that surfaced it. Approved.

[lmoresi]

This fixes an issue that was surfaced when dealing with deforming meshes but which is also quite general and is related to the surface integral code approach that we worked on recently. This should be accepted as is.