[GH-3204] GeoPandas: complete maximum-circle, dwithin-distance, and Voronoi-extent parity

[jiayuasu]

Did you read the Contributor Guide?

• Yes, I have read the Contributor Rules and Contributor Development Guide

Is this PR related to a ticket?

• Yes, and the PR name follows the format [GH-XXX] my subject.

Closes #3204

What changes were proposed in this PR?

• Implement GeoSeries.maximum_inscribed_circle with scalar, local array-like, pandas Series, and same-frame pandas-on-Spark tolerances. The implementation uses a native two-argument ST_MaximumInscribedCircle overload, preserves CRS and SRID, evaluates the circle search once per row, and handles null, empty, unsupported, and zero-extent geometries consistently.
• Extend GeoSeries.dwithin to accept local and distributed row-wise distances. Distances are paired after geometry alignment, with distributed validation for lengths and numeric types.
• Implement extend_to for GeoSeries.voronoi_polygons with native per-row Sedona expressions while accounting for the difference between GEOS extension and JTS clipping semantics.
• Keep all geometry processing distributed without collecting geometry rows to the driver or introducing Python row UDFs.

How was this patch tested?

• mvn -pl common -Dtest=FunctionsTest#maximumInscribedCircle test
• mvn -pl spark-shaded -am -DskipTests package
• Focused GeoSeries and GeoPandas parity tests: 13 passed.
• Focused Python SQL and DataFrame API tests: 7 passed.
• Native execution-plan assertions cover the absence of Python UDF nodes.
• Repository commit hooks and diff checks passed.

Did this PR include necessary documentation updates?

• Yes, I am adding a new API. I am using the current SNAPSHOT version number in v1.9.1 format.
• Yes, I have updated the SQL API documentation and GeoPandas API docstrings.