[spark] Remove upsert-key support

[JingsongLi]

Purpose

Remove the Spark upsert-key table property and its automatic INSERT INTO to MERGE INTO rewrite. It's just syntactic sugar; we shouldn't provide it because it has a lot of ambiguity with primary key tables.

This change removes the option and accessor from CoreOptions, the associated schema validation, the Spark analyzer rule and registration, and the Spark 3.2 through 4.1 tests. It also removes the SQL Upsert documentation, navigation entry, redirect, and generated configuration entry.

Existing tables that specify upsert-key will no longer receive the automatic Spark upsert rewrite. The independent PyPaimon Data Evolution upsert_by_key API is not affected.

Tests

• mvn -pl paimon-core -Pfast-build -DwildcardSuites=none -Dtest=SchemaValidationTest test
• mvn -pl paimon-spark/paimon-spark-ut -am -Pfast-build -Pspark3 -DskipTests compile
• JAVA_HOME=<jdk17> mvn -pl paimon-spark/paimon-spark-ut -am -Pfast-build -Pspark4 -DskipTests compile
• mvn -pl paimon-api,paimon-core,paimon-spark/paimon-spark-common -Pspark3 -DskipTests spotless:check

[leaves12138]

Looks good to me. I verified the complete removal of the Spark upsert-key option, analyzer rules, version-specific tests, and documentation references. SchemaValidationTest and Spark 3/Spark 4 compilation passed locally.