That seems like a great way forward, Ryan.
How should we organize around this? I feel we need to chisel out a bit more concrete plan before we hit the hackathon?
For me a key question would be: Do we focus all efforts on the first day to figure out the remapping? Or do some people move ahead with the rebinning approach outlined above and another group works on wrapping a more involved algorithm into xarray?
Also I think it is worth discussing, whether the end-product will end up as a separate package or part of xgcm.