Declare pyarrow dependency for parquet read/write

split_documents, compute_metrics, postprocessing (*_from_df) and
extract_mentions all read/write .parquet via pandas, but pandas>=2.2 does
not pull a parquet engine automatically, so these functions (and their
tests) fail at runtime without one. Add pyarrow as an explicit dependency.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: shaunpatterson

pyproject.toml

@@ -29,6 +29,7 @@ keywords = ["chunking", "nlp", "rag", "document-splitting", "evaluation"]
 dependencies = [
     "tiktoken>=0.9.0",
     "pandas>=2.2.3",
+    "pyarrow",  # parquet engine: split_documents / compute_metrics / postprocessing read+write .parquet
     "numpy",
     "tqdm>=4.67.1",
     "python-dotenv>=1.1.0",