I have a Dataflow pipeline that consists of multiple blocks that are processing heterogeneous documents (XLS, PDF etc). Each type of document is processed by a dedicated
