Add StopWordsRemover feature transformer

[devclarenjoki]

Summary

Adds a StopWordsRemover feature transformer, modeled on the existing Tokenizer, that filters stop words out of a token-array column.

• StopWordsRemover supports inputCol, outputCol, caseSensitive, and a custom stopWords list; when stopWords is unset the Spark server falls back to its default list.
• Extends ParamValue / ParamMap to carry std::vector<std::string>, emitted as a string Array literal in MlParams (toString updated to match).
• Builds the MlRelation transform (OPERATOR_TYPE_TRANSFORMER), mirroring Tokenizer.

Files

• src/ml/feature/stop_words_remover.h / .cpp — the transformer
• src/ml/param/param_map.h / .cpp — string-array param support
• tests/spark/ml/feature/stop_words_remover.cpp — integration tests (default list, custom list, chained after Tokenizer)

Testing

Integration tests run against a live Spark Connect server (docker compose up spark). Please verify locally before merge.