Try: - [x] Use mamba rather than conda - [ ] Reduce the number of variables in the datasets. `orca2_ice_pisces` has a lot of variables