[BUG] Concat of List column with scalar column results in issubclass TypeError

[oliverholworthy]

Describe the bug

Calling cudf.concat with a Series with a list dtype cudf.ListDtype("int64") and scalar dtype np.dtype("int64") raises TypeError.

Steps/Code to reproduce bug
Follow this guide http://matthewrocklin.com/blog/work/2018/02/28/minimal-bug-reports to craft a minimal bug report. This helps us reproduce the issue you're having and resolve the issue more quickly.

cudf.concat([cudf.Series([1, 2]), cudf.Series([[1, 2], [3, 4]])], axis=0, ignore_index=True)
# => Raises TypeError: issubclass() arg 1 must be a class

This is because the cudf.utils.dtypes.find_common_type function is called. Inside find_common_type the pandas function pd.api.types.is_timedelta64_dtype is called with the cudf.ListDtype("int64") which only works with numpy dtypes and not the cudf list types.

import cudf
from cudf.utils.dtypes import find_common_type
import numpy as np

find_common_type({cudf.ListDtype("int64"), np.dtype("int64")})
# Raises TypeError: issubclass() arg 1 must be a class

Expected behavior
A clear and concise description of what you expected to happen.

Either:

• catch case when concat is called with unsupported types and raise clear error message
• behaviour similar to pandas, which supports concat list and scalar values

 pd.concat([pd.Series([1, 2]), pd.Series([[1, 2], [3, 4]])], axis=0, ignore_index=True)
# =>
0         1
1         2
2    [1, 2]
3    [3, 4]
dtype: object

Environment overview (please complete the following information)

• Environment location: [Bare-metal, Docker, Cloud(specify cloud provider)]
• Method of cuDF install: [conda, Docker, or from source]
  • If method of install is [Docker], provide docker pull & docker run commands used

docker run -it --gpus=all --rm nvcr.io/nvidia/rapidsai/rapidsai:22.10-cuda11.5-runtime-ubuntu20.04-py3.8 bash

Additional context
Add any other context about the problem here.

• Came up in [BUG] Getting error when jointly encoding single-hot and multi-hot categ columns NVIDIA-Merlin/NVTabular#1639
  • Avoided this this issue by changing code to not call concat on both list an scalar series.

[oliverholworthy]

Found a more minimal example that also results in this error:

import cudf

 cudf.DataFrame({"a": [[1]]}).values
# Raises TypeError: issubclass() arg 1 must be a class