Parallel write support and example
For R3, the HPC group is interested in developing an example of how to do parallel write (e.g. using MPI). This probably requires adding some support functions to make it easier to send metadata to a central coordinator and write the chunk-metadata pages separately.