This PR adds an optimised rejection method for reading a parcel dataset. If the dataset contains the MPI decomposition boxes and start indices, they will be used to selectively read parcels, i.e. not all MPI ranks read all parcels. This increases the performance for restarts from parcel files using different number of MPI ranks as the written restart file.
This PR adds an optimised rejection method for reading a parcel dataset. If the dataset contains the MPI decomposition boxes and start indices, they will be used to selectively read parcels, i.e. not all MPI ranks read all parcels. This increases the performance for restarts from parcel files using different number of MPI ranks as the written restart file.