fix: flatten batches in PyDataFrame to ensure proper schema conversion

Author: kosiew

src/dataframe.rs

@@ -735,6 +735,7 @@ impl PyDataFrame {
 
             batches = batches
                 .into_iter()
+                .flatten()
                 .map(|record_batch| record_batch_into_schema(record_batch, &schema))
                 .collect::<Result<Vec<RecordBatch>, ArrowError>>()
                 .map_err(PyDataFusionError::from)?;