Cactus multi-block data are topologically Cartesian, but each grid point's coordinate is arbitrary.
- How would this best be handled? Should I create a UnstructuredIndex? Would "moab" be a good example to follow?
- Is it possible to combine this with AMR?