Handling non-cartesian datasets

[chrishavlin]

Would be good to add some handling of non-cartesian datasets. I'll investigate the napari side, but I suspect the solution would have to be on the yt-napari side and would involve either resampling a field on a cartesian grid. There may be some utilities in yt to help here. And there are some potential pitfalls and caveats with resampling in cartesian, but I think we could come up with a reasonable solution.

[chrishavlin]

was playing with this briefly today: can actually extract a regularized region with ds.r for non-cartesian geometries and load this in linear coordinates. Would need to add an extra interpolation/transformation to view in true cartesian coordinates though.