[BUG]: cuspatial API example notebook throws error ndarray attribute error when running pip

[taureandyernv]

Version

23.02 nightly

On which installation method(s) does this occur?

Docker

Describe the issue

when running PIP, I get an Attribute error: AttributeError: 'ndarray' object has no attribute 'index'.

Minimum reproducible example

import cuspatial
import cupy
import geopandas
host_dataframe = geopandas.read_file(geopandas.datasets.get_path("naturalearth_lowres"))
gpu_dataframe = cuspatial.from_geopandas(host_dataframe)
x_points = (cupy.random.random(10000000) - 0.5) * 360
y_points = (cupy.random.random(10000000) - 0.5) * 180

short_dataframe = gpu_dataframe.iloc[0:32]
geometry = short_dataframe['geometry']
points_in_polygon = cuspatial.point_in_polygon(
    x_points,
    y_points,
    geometry.polygons.geometry_offset[0:31],
    geometry.polygons.ring_offset,
    geometry.polygons.x,
    geometry.polygons.y
)

Relevant log output

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
/tmp/ipykernel_4178/1403701831.py in <module>
      6 short_dataframe = gpu_dataframe.iloc[0:32]
      7 geometry = short_dataframe['geometry']
----> 8 points_in_polygon = cuspatial.point_in_polygon(
      9     x_points,
     10     y_points,

/opt/conda/envs/rapids/lib/python3.10/site-packages/cuspatial/core/spatial/join.py in point_in_polygon(test_points_x, test_points_y, poly_offsets, poly_ring_offsets, poly_points_x, poly_points_y)
    116         {name: col.astype("bool") for name, col in result._data.items()}
    117     )
--> 118     result.columns = [x for x in list(reversed(poly_offsets.index))]
    119     result = result[list(reversed(result.columns))]
    120     return result

AttributeError: 'ndarray' object has no attribute 'index'

Environment details

This happens in all docker containers for the docker nightly builds as of 2/6

Other/Misc.

No response

[isVoid]

Though not ideal, this will be temporarily fixed by #906

[harrism]

Can you explain why the fix is not ideal, @isVoid ?

[isVoid]

cuspatial.point_in_polygon documentation doesn't detail the requirement of the input arguments. It seems as though the inputs can just be cupy arrays, except for poly_offsets - .index property is accessed so this would need to be a series instead. I think some refactoring and additional documentation to the API is needed.