BUG: Fix display with nested arrays

Currently, the following will fail to display:
```python
import xarray as xr
import numpy as np
x = np.empty((2, 2), dtype=object)
for i in range(2):
    for j in range(2):
        x[i, j] = np.zeros(2)  # Set to 1D array of size 2
ds = xr.DataArray(x)
ds
```

Whenever there are `size==1` arrays, it currently does work:
```python
import xarray as xr
import numpy as np
x = np.empty((2, 2), dtype=object)
for i in range(2):
    for j in range(2):
        x[i, j] = np.zeros((1, 1, 1))  # Set to 3D array of size 1
ds = xr.DataArray(x)
ds

For context, I am running into this issue in https://github.com/pipefunc/pipefunc where one can put the resulting objects into `xarray.Dataset`s.
For example, see the docs here https://pipefunc.readthedocs.io/en/latest/examples/physics-simulation/ (search for `xarray.Dataset`).

Author: basnijholt

xarray/core/formatting.py

@@ -184,7 +184,7 @@ def format_item(x, timedelta_format=None, quote_strings=True):
         if hasattr(x, "dtype"):
             x = x.item()
         return repr(x) if quote_strings else x
-    elif hasattr(x, "dtype") and np.issubdtype(x.dtype, np.floating):
+    elif hasattr(x, "dtype") and np.issubdtype(x.dtype, np.floating) and x.size == 1:
         return f"{x.item():.4}"
     else:
         return str(x)

xarray/tests/test_formatting.py

@@ -101,6 +101,8 @@ def test_format_item(self) -> None:
             (np.float16(1.1234), "1.123"),
             (np.float32(1.0111111), "1.011"),
             (np.float64(22.222222), "22.22"),
+            (np.zeros((1, 1)), "array([[0.]])"),
+            (np.zeros(2), "array([0., 0.])"),
         ]
         for item, expected in cases:
             actual = formatting.format_item(item)