Feature request: Handle distributed catalogues

[MatthieuSchaller]

It's much much more efficient to run VR without parallel-hdf5. Being able to handle such distributed catalogs with the python package would be grand.

[bwvdnbro]

This would probably require quite a rewrite of the way the data is read from the catalogue.

We could have a script that generates a virtual file that can act as a single-file entry point for the python package?