TST: Retyping of categorical column with NaN (#60112)

* consistent name usage

* changed to numpy array of integers

* Remove redundant assert

Co-authored-by: Matthew Roeschke <10647082+mroeschke@users.noreply.github.com>

---------

Co-authored-by: Matthew Roeschke <10647082+mroeschke@users.noreply.github.com>

Author: SpoopyPillow

pandas/tests/dtypes/test_dtypes.py

@@ -1243,3 +1243,12 @@ def test_loc_setitem_empty_labels_no_dtype_conversion():
 
     assert df.a.dtype == "int64"
     tm.assert_frame_equal(df, expected)
+
+
+def test_categorical_nan_no_dtype_conversion():
+    # GH 43996
+
+    df = pd.DataFrame({"a": Categorical([np.nan], [1]), "b": [1]})
+    expected = pd.DataFrame({"a": Categorical([1], [1]), "b": [1]})
+    df.loc[0, "a"] = np.array([1])
+    tm.assert_frame_equal(df, expected)