Implement datatree_to_icechunk such that the previously added test passes

[maxrjones]

No description provided.

[maxrjones]

The new functionality for this would go in https://github.com/zarr-developers/VirtualiZarr/blob/main/virtualizarr/writers/icechunk.py. I think it may be possible to implement this before the xarray DataTree <-> Zarr Python 3 issues are resolved if datatree_to_icechunk iterates over the nodes and calls write_variables_to_icechunk_group on each node.

[chuckwondo]

The new functionality for this would go in https://github.com/zarr-developers/VirtualiZarr/blob/main/virtualizarr/writers/icechunk.py. I think it may be possible to implement this before the xarray DataTree <-> Zarr Python 3 issues are resolved if datatree_to_icechunk iterates over the nodes and calls write_variables_to_icechunk_group on each node.

@maxrjones, when you say "nodes," do you mean datasets? If so, does that mean each time we call write_variables_to_icechunk_group, is the group the absolute path of the dataset within the tree?

[maxrjones]

If you look at https://docs.xarray.dev/en/stable/getting-started-guide/quick-overview.html#datatrees I'm referring to what that section calls "groups". While those docs say "you can think of it (a datatree) as a recursive dict of Dataset objects", I think it's important for this issue to recognize that a node/group is not exactly a dataset. You can call DataTree[<node-name>].ds to get a view of the contained dataset to DataTree[<node-name>].to_dataset() to get a copy, but the nodes implement different methods than datasets which is why they are different.