Collect errors during PRE retrieval #51
Description
This is technically what the reference does, but I am not sure how useful it is - if some node are not responsive, this will be handled by NetworkClient. If nodes send high-level incorrect responses, perhaps we should just report back to NetworkClient so that those nodes are flagged?
What should Proxy do - would the user be interested in node errors? Can we detect all the errors at the request stage, or can some only be detected by the final recipient?