Apply metadata to keys before returning in Frame._encode (#8560)

Fixes #7365 

Applies column metadata to the output columns of `keys` in `Frame._encode`; skipping this step meant that the output of `DataFrame.unstack` would not have the expected metadata for index columns:

```python
import pandas as pd
import cudf

pdf = pd.DataFrame(
    {
        "foo": ["one", "one", "one", "two", "two", "two"],
        "bar": pd.Categorical(["A", "B", "C", "A", "B", "C"]),
        "baz": [1, 2, 3, 4, 5, 6],
        "zoo": ["x", "y", "z", "q", "w", "t"],
    }).set_index(["foo", "bar", "baz"])
gdf = cudf.from_pandas(pdf)

pdf.unstack("baz")
         zoo                         
baz        1    2    3    4    5    6
foo bar                              
one A      x  NaN  NaN  NaN  NaN  NaN
    B    NaN    y  NaN  NaN  NaN  NaN
    C    NaN  NaN    z  NaN  NaN  NaN
two A    NaN  NaN  NaN    q  NaN  NaN
    B    NaN  NaN  NaN  NaN    w  NaN
    C    NaN  NaN  NaN  NaN  NaN    t

gdf.unstack("baz")
          zoo                              
baz         1     2     3     4     5     6
foo bar                                    
one 0       x  <NA>  <NA>  <NA>  <NA>  <NA>
    1    <NA>     y  <NA>  <NA>  <NA>  <NA>
    2    <NA>  <NA>     z  <NA>  <NA>  <NA>
two 0    <NA>  <NA>  <NA>     q  <NA>  <NA>
    1    <NA>  <NA>  <NA>  <NA>     w  <NA>
    2    <NA>  <NA>  <NA>  <NA>  <NA>     t
```

Authors:
  - Charles Blackmon-Luca (https://github.com/charlesbluca)

Approvers:
  - GALI PREM SAGAR (https://github.com/galipremsagar)

URL: #8560

python/cudf/cudf/core/frame.py

@@ -3251,6 +3251,11 @@ def _split(self, splits, keep_index=True):
     def _encode(self):
         keys, indices = libcudf.transform.table_encode(self)
         keys = self.__class__._from_table(keys)
+        for col in keys._data:
+            keys._data[col] = keys._data[col]._with_type_metadata(
+                self._data[col].dtype
+            )
+
         return keys, indices
 
     def _reindex(

python/cudf/cudf/tests/test_reshape.py

@@ -393,15 +393,35 @@ def test_pivot_multi_values():
     "level",
     [
         0,
-        1,
+        pytest.param(
+            1,
+            marks=pytest.mark.xfail(
+                reason="Categorical column indexes not supported"
+            ),
+        ),
         2,
         "foo",
-        "bar",
+        pytest.param(
+            "bar",
+            marks=pytest.mark.xfail(
+                reason="Categorical column indexes not supported"
+            ),
+        ),
         "baz",
         [],
-        [0, 1],
+        pytest.param(
+            [0, 1],
+            marks=pytest.mark.xfail(
+                reason="Categorical column indexes not supported"
+            ),
+        ),
         ["foo"],
-        ["foo", "bar"],
+        pytest.param(
+            ["foo", "bar"],
+            marks=pytest.mark.xfail(
+                reason="Categorical column indexes not supported"
+            ),
+        ),
         pytest.param(
             [0, 1, 2],
             marks=pytest.mark.xfail(reason="Pandas behaviour unclear"),
@@ -416,7 +436,7 @@ def test_unstack_multiindex(level):
     pdf = pd.DataFrame(
         {
             "foo": ["one", "one", "one", "two", "two", "two"],
-            "bar": ["A", "B", "C", "A", "B", "C"],
+            "bar": pd.Categorical(["A", "B", "C", "A", "B", "C"]),
             "baz": [1, 2, 3, 4, 5, 6],
             "zoo": ["x", "y", "z", "q", "w", "t"],
         }
@@ -436,6 +456,12 @@ def test_unstack_multiindex(level):
     [
         pd.Index(range(0, 5), name=None),
         pd.Index(range(0, 5), name="row_index"),
+        pytest.param(
+            pd.CategoricalIndex(["d", "e", "f", "g", "h"]),
+            marks=pytest.mark.xfail(
+                reason="Categorical column indexes not supported"
+            ),
+        ),
     ],
 )
 @pytest.mark.parametrize(