How to Work Around PyArrow's 'from_pandas' Crash with Mixed Dtypes and Custom Type Conversion

Efficient Data Import: Reading Parquet Files in Chunks and Inserting into DuckDB