Deep copying a datatree somehow duplicates data onto child nodes

[TomNicholas]

What happened?

Deep copying a DataTree object somehow duplicates data onto a child node (!)

What did you expect to happen?

Copying to return an identical subtree.

Minimal Complete Verifiable Example

In [15]: dt = DataTree(data=xr.Dataset(coords={'x': ('x', [0, 1])}))

In [16]: print(dt)
<xarray.DataTree>
Group: /
    Dimensions:  (x: 2)
    Coordinates:
      * x        (x) int64 16B 0 1
    Data variables:
        *empty*

In [17]: dt['c'] = DataTree()

In [18]: print(dt)
<xarray.DataTree>
Group: /
│   Dimensions:  (x: 2)
│   Coordinates:
│     * x        (x) int64 16B 0 1
│   Data variables:
│       *empty*
└── Group: /c

In [19]: dt.copy(deep=True)
Out[19]: 
<xarray.DataTree>
Group: /
│   Dimensions:  (x: 2)
│   Coordinates:
│     * x        (x) int64 16B 0 1
│   Data variables:
│       *empty*
└── Group: /c
        Dimensions:  (x: 2)
        Coordinates:
          * x        (x) int64 16B 0 1
        Data variables:
            *empty*

Anything else we need to know?

No response

Environment

main

[TomNicholas]

Wait I'm being an idiot - the coordinates appearing in both groups is expected behaviour now that we have coordinate inheritance.

What's weird is that they don't appear in the child node when we print(dt)! This implies some bug with coordinate inheritance / handling of empty nodes (cc @shoyer).

[shoyer]

I think I know what's going on... stay tuned for a fix