Check num_children() == 0 in Column.from_column_view (#17193)

This fixes a bug where `Column.from_column_view` is not verifying the existence of a string column's offsets child column prior to accessing it, resulting in a segmentation fault when passing a `column_view` from `Column.view()` to `Column.from_column_view(...)`.

The issue can be reproduced with:
```
import cudf
from cudf.core.column.column import as_column
df = cudf.DataFrame({'a': cudf.Series([[]], dtype=cudf.core.dtypes.ListDtype('string'))})
s = df['a']
col = as_column(s)
col2 = cudf._lib.column.Column.back_and_forth(col)
print(col)
print(col2)
```
where `back_and_forth` is defined as:
```
    @staticmethod
    def back_and_forth(Column input_column):
        cdef column_view input_column_view = input_column.view()
        return Column.from_column_view(input_column_view, input_column)
```
I don't have the expertise to write the appropriate tests for this without introducing the `back_and_forth` function as an API, which seems undesirable.

Authors:
  - Christopher Harris (https://github.com/cwharris)

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

URL: #17193

python/cudf/cudf/_lib/column.pyx

@@ -688,15 +688,18 @@ cdef class Column:
         # special case for string column
         is_string_column = (cv.type().id() == libcudf_types.type_id.STRING)
         if is_string_column:
-            # get the size from offset child column (device to host copy)
-            offsets_column_index = 0
-            offset_child_column = cv.child(offsets_column_index)
-            if offset_child_column.size() == 0:
+            if cv.num_children() == 0:
                 base_nbytes = 0
             else:
-                chars_size = get_element(
-                    offset_child_column, offset_child_column.size()-1).value
-                base_nbytes = chars_size
+                # get the size from offset child column (device to host copy)
+                offsets_column_index = 0
+                offset_child_column = cv.child(offsets_column_index)
+                if offset_child_column.size() == 0:
+                    base_nbytes = 0
+                else:
+                    chars_size = get_element(
+                        offset_child_column, offset_child_column.size()-1).value
+                    base_nbytes = chars_size
 
         if data_ptr:
             if data_owner is None: