Fix Series.groupby.shift with a MultiIndex (#15098)

closes #15087
closes #11259

(The typing annotation is incorrect, but I guess there needs to be a check somewhere to make `_copy_type_metadata` stricter)

Authors:
  - Matthew Roeschke (https://github.com/mroeschke)

Approvers:
  - Bradley Dice (https://github.com/bdice)

URL: #15098

python/cudf/cudf/core/multiindex.py

@@ -2037,7 +2037,8 @@ def _copy_type_metadata(
         self: MultiIndex, other: MultiIndex, *, override_dtypes=None
     ) -> MultiIndex:
         res = super()._copy_type_metadata(other)
-        res._names = other._names
+        if isinstance(other, MultiIndex):
+            res._names = other._names
         return res
 
     @_cudf_nvtx_annotate

python/cudf/cudf/tests/test_groupby.py

@@ -3308,7 +3308,6 @@ def test_groupby_pct_change(data, gkey, periods, fill_method):
     assert_eq(expected, actual)
 
 
-@pytest.mark.xfail(reason="https://github.com/rapidsai/cudf/issues/11259")
 @pytest.mark.parametrize("periods", [-5, 5])
 def test_groupby_pct_change_multiindex_dataframe(periods):
     gdf = cudf.DataFrame(
@@ -3812,3 +3811,13 @@ def test_groupby_internal_groups_empty(gdf):
     gb = gdf.groupby("y")._groupby
     _, _, grouped_vals = gb.groups([])
     assert grouped_vals == []
+
+
+def test_groupby_shift_series_multiindex():
+    idx = cudf.MultiIndex.from_tuples(
+        [("a", 1), ("a", 2), ("b", 1), ("b", 2)], names=["f", "s"]
+    )
+    ser = Series(range(4), index=idx)
+    result = ser.groupby(level=0).shift(1)
+    expected = ser.to_pandas().groupby(level=0).shift(1)
+    assert_eq(expected, result)