Fix broken hausdorff test

[isVoid]

This PR revises the gold result of hausdorff empty input test to address
an upstream change in cudf.

Description

Checklist

• I am familiar with the Contributing Guidelines.
• New or existing tests cover these changes.
• The documentation is up to date with these changes.

[vyasr]

Fix seems fine to me, although I wonder if maybe it should be changed the other way in the long term i.e. have cuspatial return an empty RangeIndex rather than force cudf to produce an empty Index.

[harrism]

Indeed that's what chatGPT thinks. https://chat.openai.com/share/43a1b380-e356-4b5e-8c87-08b07003435b

How hard would it be to fix cuSpatial to match cuDF here?

[isVoid]

Maybe not that hard - probably need one more argument for _from_columns call.

[isVoid]

Fundamentally these two calls are different:

In [4]: print(cudf.DataFrame([]).columns)
RangeIndex(start=0, stop=0, step=1)

In [5]: print(cudf.DataFrame._from_columns([], range(0)).columns)
Index([], dtype='object')

One is the public API that cuspatial generate the gold result from, and the other is the internal factory function that cuspatial depends upon. I think the proper way forward is that cuspatial shouldn't rely on using internal function from cuDF and attempt to reconstruct the result using public API.

[isVoid]

Ooops, I looked at the wrong place - empty input has a shortcut that directly return an empty dataframe. So the actual cause of error is:

In [3]: cudf.DataFrame([]).columns
Out[3]: RangeIndex(start=0, stop=0, step=1)

In [4]: cudf.DataFrame().columns
Out[4]: Index([], dtype='object')

[isVoid]

Ok I think I prefer this fix - not having to fix the golden result seems a better way forward.

[vyasr]

I agree I prefer this fix. Thanks for indulging my question!