Fix an issue with fetching NA from a TimedeltaColumn (#13853)

Fixes: #13851 

This PR fixes a mistake in `_get_np_scalar_from_timedelta64`, which has to return `cudf.NA` instead of `None`.

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

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

URL: #13853

python/cudf/cudf/_lib/scalar.pyx

@@ -1,4 +1,4 @@
-# Copyright (c) 2020-2022, NVIDIA CORPORATION.
+# Copyright (c) 2020-2023, NVIDIA CORPORATION.
 
 cimport cython
 
@@ -536,7 +536,7 @@ cdef _get_np_scalar_from_timedelta64(unique_ptr[scalar]& s):
     cdef scalar* s_ptr = s.get()
 
     if not s_ptr[0].is_valid():
-        return None
+        return NA
 
     cdef libcudf_types.data_type cdtype = s_ptr[0].type()
 

python/cudf/cudf/tests/test_timedelta.py

@@ -1437,3 +1437,8 @@ def test_timdelta_binop_tz_timestamp(op):
     date_tz_scalar = datetime.datetime.now(datetime.timezone.utc)
     with pytest.raises(NotImplementedError):
         op(s, date_tz_scalar)
+
+
+def test_timedelta_getitem_na():
+    s = cudf.Series([1, 2, None, 3], dtype="timedelta64[ns]")
+    assert s[2] is cudf.NA