[GH-3260] Add ST_EqualsIdentical across Sedona SQL engines

[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.

Part of #3260.

The serialization prerequisites #3265 and #3268 have merged into master.

What changes were proposed in this PR?

• Add a common ST_EqualsIdentical implementation that compares exact geometry type, nested structure, component and coordinate order, dimensionality, and every XY/Z/M ordinate.
• Treat corresponding NaN ordinates as equal, treat positive and negative zero as equal, and ignore SRID and other geometry metadata.
• Register and expose the predicate through:
  • Spark SQL, DataFrame APIs, Spark Connect, and the Python Spark SQL API
  • Flink SQL
  • Snowflake WKB and native GEOMETRY UDF paths
• Add SQL documentation for Spark, Flink, and Snowflake, including each engine's stored-geometry dimensionality boundary and practical round-trip validation examples.

This is the second PR in the #3260 stack and is now based directly on master. Its two serialization prerequisites are merged:

• [GH-3260] Preserve M and ZM in Python GeometryUDT serialization #3265 preserves M and ZM when Shapely geometries cross the Python GeometryUDT boundary.
• [GH-3267] Preserve recoverable geometry dimensions #3268 preserves every recoverable coordinate layout when geometries cross Sedona's compact Java SerDe boundary, and rejects mixed layouts that its single-layout container encoding cannot represent losslessly.

ST_EqualsIdentical compares the geometry representation it receives; it cannot reconstruct dimensional information already discarded by an upstream representation. JTS 1.20's default coordinate-sequence factory still makes ordinary XY indistinguishable from declared XYZ whose Z ordinates are all NaN, while Snowflake native GEOMETRY stores normalized 2D coordinates. The engine-specific documentation calls out these boundaries.

The final stacked PR is #3262, which adds distributed GeoPandas geom_equals_identical on top of this predicate.

How was this patch tested?

• Full common test suite: 1,304 tests passed.
• Spark predicate suite: 37 tests passed.
• Spark DataFrame API suite: 261 tests passed.
• Flink predicate suite: 24 tests passed.
• Snowflake DDL suite: 3 tests passed; both modified Snowflake tester suites compiled successfully.
• Regression coverage includes XY/Z/M/ZM distinctions, NaN and infinity handling, represented typed empties, component and ring ordering, SRID independence, null propagation, API argument validation, and SerDe-preserved dimensional layouts.
• The repository commit-hook suite, Spotless, documentation build, Markdown linting, and git diff --check passed.

Did this PR include necessary documentation updates?

• Yes, I am adding a new API. I am using the current SNAPSHOT version, v2.0.0.
• Yes, I added Spark, Flink, and Snowflake SQL documentation.