Handle sparse GeoSeries better and add expand_values to convert sparse to dense.

[thomcom]

Fixes #1142

Description

When a GeoSeries is sliced, or through careful construction methods, a sparse GeoSeries can be created:

series = cuspatial.GeoSeries([Point(0, 0)])
sparse_series = series[[0, 0]]
sparse_series

0    POINT (0.00000 0.00000)
0    POINT (0.00000 0.00000)
dtype: geometry

The above series has the appearance of containing two Points, but actually contains a single Point which is indexed twice. pairwise_linestring_intersection depends on a pairwise comparison of dense LineStrings. This PR adds GeoSeries.expand_values() which fills out a sparse GeoSeries prior to calling pairwise_linestring_intersection.

Checklist

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

[thomcom]

This PR extends on the functionality in https://github.com/rapidsai/cuspatial/blob/branch-23.06/python/cuspatial/cuspatial/core/geoseries.py#L256 by allowing the features in a sparse mixed geometry column to be expanded. I don't think we need this right now however and am tempted to just close this PR as unneeded. For now, geoseries.lines.column() implements the required functionality.

[thomcom]

Unnecessary.