Easier data access for parallel VTK formats
It would be useful for users if there was an easier method to fetch data from parallel VTK formats. Currently, a user will need to step through all the pieces of a multipiece dataset, and fetch the arrays individually. Then there are also issues like ghost cells and points that need to be taken into account.
So it would be good to have a method to fetch a cell or point data array that
- contains all data over all partitions
- returns data in a consistent manner
- takes into account ghost cells or ghost points
Another way to phrase this is "give me the data as if this was a serial dataset". We've discussed it with Kitware, and some method in the Numpy VTK integration sounds like a natural fit.