Interaction of Hydra with MPI, in particular mpi4py

[philippweder]

I am using hydra in combination with mpi4py. More precisely, as a part of my FEM pipeline based on FEniCSx, I load a mesh using dolfinx.read_from_msh.
Without hydra, everything works fine. However, my main script that uses hydra suddenly starts throwing errors of the following type:

Invalid rank, error stack:
internal_Issend(60788): MPI_Issend(buf=0x7f80547eb421, count=1, MPI_BYTE, 1, 1, comm=0x84000005, request=0x7f80547eb414) failed
internal_Issend(60749): Invalid rank has value 1 but must be nonnegative and less than 1
Abort(339346182) on node 0 (rank 0 in comm 416): application called MPI_Abort(comm=0x84000005, 339346182) - process 0

This suggests that the internal MPI configuration somehow got changed by hydra such that my code now suddenly assumes that there is more than one process. Where can I handle that in hydra?