Chunked Query Results in the DuckDB Java Driver

DuckDB is a columnar, vectorized database. Every operator inside the engine works on data chunks: batches of column vectors, 2,048 rows at a time. That is a big part of why DuckDB is fast: the engine amortizes interpretation overhead over thousands of values instead of paying it once per value.