Fix sortby raising KeyError when passed a tuple of dimension names

[thodson-usgs]

Summary

• Fixes da.sortby(da.dims) raising KeyError when da.dims is a tuple (e.g. ('x', 'y'))
• Tuples are now treated as iterables of variable names, consistent with the str | Iterable[Hashable] convention established in dimensions: type as str | Iterable[Hashable]? #6142

Closes #4821

Usage example

>>> import xarray as xr
>>> da = xr.DataArray(
...     [[1, 2], [3, 4], [5, 6]],
...     coords={"x": ["c", "b", "a"], "y": [1, 0]},
... )
>>> da.sortby(da.dims)  # da.dims is the tuple ("x", "y")
<xarray.DataArray (x: 3, y: 2)> Size: 48B
array([[6, 5],
       [4, 3],
       [2, 1]])
Coordinates:
  * x        (x) <U1 12B 'a' 'b' 'c'
  * y        (y) int64 16B 0 1

Test plan

• da.sortby(da.dims) works when da.dims is a tuple of dimension names
• ds.sortby(("x", "y")) sorts by both dimensions
• Existing sortby tests pass (string, list, DataArray, descending, NaN handling)
• whats-new.rst entry added

[This is Claude Code on behalf of Timothy Hodson]

[headtr1ck]

Could you update the type hints as well?
Should probably be

str | DataArray | Iterable[Hashable | DataArray] | Callable[[Self], str | DataArray | Iterable[Hashable | DataArray]]

Same with the docstring and the DataArray method.

Thanks!