Support cupy and numpy array types in `flatten_list_column_values`

[oliverholworthy]

Follow-up to #244

This PR adds support for cupy and numpy array types in flatten_list_column_values.

The motivation for this comes from:

• Normalize op in NVTabular is expected to work on dictionaries of cupy arrays. It calls flatten_list_column_values if the array is list-like.
  • If this is passed a list-like cupy array, the normalize operator expects to be able to flatten this value.
• pd.api.types.is_list_like return value for cupy array's changes depending on version of the pandas package.
  • in earlier versions of pandas (e.g. 1.3.5)
    • pd.api.types.is_list_like(cupy.array(1)) # => True
  • in more recent version of pandas (e.g. 1.5.3)
  • pd.api.types.is_list_like(cupy.array(1)) # => False