Support Tripolar / Unstructured grids

[max-anu]

A lot of data will not be on a lat/lon grid. We should add support for other common grids such as triangular/hexagonal grids.
This was brought up in a webinar.

[OKaluza]

Thanks for starting the discussion, I think if we want to support other grids this will need to be done case-by-case, with specific examples. Any triangular mesh is easy as you can send triangles directly to lavavu, hexagonal will have to be split into 4 triangles but that's also reasonably easy.
Regarding how common this is - I haven't come across much other than lon/lat coord systems in data sets I've plotted so far in my dealings with climate data so reiterate the need for examples. Other geo coord systems can easily be converted to lon/lat and I'd argue that conversion isn't in scope for us, we just need to define clearly the input data specifications.
To plot data very close to poles it makes sense to avoid lat/lon so might be nice to convert from whatever grid is in use in these cases directly to 3d coords.
I'm definitely open to writing convenience functions for common use cases such as this, and handling tripolar grids but I think each use case should be an issue on its own. Until then we can keep this open as a place to collect examples and decide if to support them.

[rbeucher]

Thanks @max-anu. It's a good suggestion. Not super high on the priority list but definitely something to look into.
ISSM models have unstructured grids so if @headmetal has some outputs that could be a good example.
MOM and CICE have tripolar grid #47

[headmetal]

Thanks @max-anu , @rbeucher

@lawrenceabird is currently working on developing capability to ingest/convert unstructured ISSM output grids to structured grids following ISMIP standards in pyISSM. Happy to share what we are doing as this matures.