Deepak Cherian

I suspect the issue is in the metpy function which might be converting to numpy. We'll need to see the whole traceback to be sure

I think a better solution here is to explicitly model the periodic grid. I have a different attempt here: https://github.com/earth-mover/xpublish-tiles/blob/main/src/xpublish_tiles/grids.py which could be broken out to its own package.

Note to myself: this needs some harmonizing of `_DUMMY_AXIS` is `-2` and the fact that the `k`-axis is `0`

I wrote some helpers to "cache" the fractional coverage [here](https://rasterix.readthedocs.io/en/latest/rasterize/exactextract.html). Going from there to the statistics i have not done yet; but the mean should just be `xr.dot(frac, dataset)` I...