[SPARK-52243][CONNECT] Add NERF support for schema-related InvalidPlanInput errors

[heyihong]

What changes were proposed in this pull request?

This PR adds NERF (New Error Framework) support for schema-related InvalidPlanInput errors in Spark Connect. The changes include:

1. Added new error conditions in error-conditions.json for schema validation:

   • INVALID_OUTPUT_SCHEMA_TYPE_FOR_TRANSFORM_WITH_STATE_IN_PANDAS_NON_STRUCT
   • INVALID_PYTHON_UDTF_RETURN_TYPE_NON_STRUCT
   • INVALID_SCHEMA_TYPE_NON_STRUCT
   • INVALID_STATE_SCHEMA_TYPE_FOR_FLAT_MAP_GROUPS_WITH_STATE_NON_STRUCT

2. Refactored error handling in InvalidInputErrors.scala to use the new NERF framework:

   • Added helper function invalidPlanInput for consistent error message generation
   • Updated schema validation error methods to use NERF error conditions
   • Made quoteByDefault method accessible to other packages

3. Added a test suite InvalidInputErrorsSuite.scala to verify error handling

Why are the changes needed?

These changes are needed to:

1. Standardize error reporting across Spark Connect using the NERF framework
2. Improve error messages with better parameterization and consistency
3. Ensure proper SQL state codes are associated with schema-related errors
4. Provide clearer error messages for users when schema validation fails

Does this PR introduce any user-facing change?

No

How was this patch tested?

build/sbt "connect/testOnly *InvalidInputErrorsSuite"

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

Generated-by: Cursor 0.50.5 (Universal)

[heyihong]

@HyukjinKwon @cloud-fan @hvanhovell

[cloud-fan]

Shall we use this error condition everywhere? All of those features need to specify a schema, and INVALID_SCHEMA_TYPE_NON_STRUCT is general enough that fit all of them.

[cloud-fan]

To indicate where the error happened, error context is a better place.