In an MPI distributed setting, a `vtkOpenFOAMReader` property update using a `StreamingDemandDriverPipeline` would only trigger the `RequestData` for rank 0. For reconstructed cases, vtkPOpenFOAMReader only uses rank 0 to do the work, but would still wait for all processes to go through RequestData, which never happens, causing an infinite hang. By setting `CAN_HANDLE_PIECE_REQUEST` in every case, we ensure that all processes go through RequestData, even if they have no actual work to do, so the rank 0 can go across the barrier.