[SPARK-54532][PYTHON] Add sqlstate for a set of PySparkException error classes

[khakhlyuk]

What changes were proposed in this pull request?

Following #53241 I am adding sql states for a set of PySparkException error classes.

I chose the following sql states mappings:

• 42K09 - "Data type mismatch"
  • a long list of error classes like NOT_LIST_OR_STR_OR_TUPLE, quite self-explanatory
• 0A000 - "feature not supported"
  • ATTRIBUTE_NOT_SUPPORTED, CLASSIC_OPERATION_NOT_SUPPORTED_ON_DF, JVM_ATTRIBUTE_NOT_SUPPORTED, NOT_IMPLEMENTED, ONLY_SUPPORTED_WITH_SPARK_CONNECT
• 08003 - "connection does not exist"
  • NO_ACTIVE_SESSION, NO_ACTIVE_OR_DEFAULT_SESSION
  • The 08003 SQL state is a standard ANSI/ISO state meaning "connection does not exist". A Spark session is conceptually similar to a database connection. If there is no active session, a connection doesn't exist.
• 08002 - "connection name in use" (spark) / "connection already exists" (IBM DB2)
  • SESSION_ALREADY_EXIST
  • "Connection already exists" would map perfectly to the error class, "connection name in use" sounds a bit confusing but semantically 08002 is still the best sql state for SESSION_ALREADY_EXIST, especially if we choose 08003 for NO_ACTIVE_SESSION.

Why are the changes needed?

sql states allow to better categorize and understand exceptions coming from pyspark

Does this PR introduce any user-facing change?

Users will be able to read the newly populated sql states using e.getSqlState()

How was this patch tested?

NA, mapping changes.

Was this patch authored or co-authored using generative AI tooling?

No

[allisonwang-db]

cc @srielau

[HyukjinKwon]

Merged to master.