Function drain_data_buffer_to_record_batches 

fn drain_data_buffer_to_record_batches(
    schema: SchemaRef,
    buffer: &mut DataBuffer,
    pk_weights: Option<&[u16]>,
    dedup: bool,
    replace_pk_index: bool,
) -> Result<RecordBatch>

Converts DataBuffer to record batches, with rows sorted according to pk_weights. dedup: whether to true to remove the duplicated rows inside DataBuffer. replace_pk_index: whether to replace the pk_index values with corresponding pk weight.