[GH-2565] Fix NULL handling for various aggregation functions in SedonaSpark

[Kontinuation]

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-2565] my subject. Closes Aggregation functions in SedonaSpark throw exception when processing NULL values #2565

What changes were proposed in this PR?

This PR refactors several spatial aggregate functions to fix NULL handling. Impacted functions are

• ST_Envelope_Aggr: fixed null handling, simplified the implementation by switching to lightweight intermediate data structure
• ST_Union_Aggr: fixed null handling
• ST_Intersection_Aggr: fixed null handling
• ST_Collect_Agg: return null instead of empty geometry collection when all the inputs are null. This is theoretically a breaking change, but it is a niche case so hopefully it won't impact many users.

How was this patch tested?

Added new tests and tested locally using a standalone cluster to make sure that we are not damaging the serde in distributed environment.

Did this PR include necessary documentation updates?

• No, this PR does not affect any public API so no need to change the documentation.

[Kontinuation]

The Python build extension failure is probably caused by the newly released PySpark 4.1.0 not playing well with Windows. See apache/spark#50466