fix: unable to create cutout when dataset vars are ordered differently

[veech]

Description

When attempting to create a limited area cutout using a dataset definition like this:

dataset: 
  cutout:
    - era5-ml88-conus-0p25-2020010100-2020013123-1h-v2.zarr
    - join:
        - era5-ml137t49-geopotential-0p25-2020010100-2020013123-1h-v1.zarr
        - era5-ml137t49-sp-0p25-2020010100-2020013123-1h-v1.zarr
        - era5-ml137t49-temperature-0p25-2020010100-2020013123-1h-v1.zarr
        - era5-ml137t49-uwind-0p25-2020010100-2020013123-1h-v1.zarr
        - era5-ml137t49-vorticity-0p25-2020010100-2020013123-1h-v1.zarr
        - era5-ml137t49-vwind-0p25-2g020010100-2020013123-1h-v1.zarr
        - era5-sfc-all-0p25-2020010100-2020013123-1h-v1.zarr

I would get the error:

ValueError: Incompatible variables: 

This comes from here.

Even though the datasets have the same variables, it must be that when I join the boundary dataset, the variables are in a different order.

[FussyDuck]

All committers have signed the CLA.

[floriankrb]

If the order of the variables is not correct, open_dataset rightfully complains that the data can not be merged

To avoid this, the variables can be reordered using select: and give a list : https://anemoi.readthedocs.io/projects/datasets/en/latest/datasets/using/selecting.html

It can be tested with : ds = open_dataset(dataset, select=["2t", "tp"])

--
Alternatively, you may want to try https://anemoi.readthedocs.io/projects/datasets/en/latest/datasets/using/matching.html#using-matching to adjust automatically the variables when doing a concat ds = open_dataset(concat=[dataset1, dataset2], adjust="variables")
But I don't think it applies directly to your case, it seems that each dataset contains only one variable.