Henk,
If the RFC editor will find these errors, then I suggest to move their review to an earlier stage and/or instruct the IESG to focus on the content, not editorial issues.
All end of the process reviews (IESG, LC, directorate, etc) already focus on content. As several people read a document, some editorial issues are bound to come up and they get reported. In fact, I think we get plenty of them for each document. However, they are not blocking. From an AD perspective they would result in a Comment, not a Discuss. The authors are informed of the issues, but there is no requirement to fix such problems.
By the way, we had an experiment couple of years ago where RFC Editor edits were performed before IETF LC. My personal sense of it was that it wasn't all that useful for a couple of reasons. First of all, if we are just talking about editorial issues, they could easily be done in the RFC Editor stage, too. Secondly, we realized that a number of changes were still being made at and after IETF LC, so any editorial cleanup would in any case have to be done later for the changed parts. Finally, the RFC Editor process speed increased significantly, so there was no reason to attempt to parallelize the editing process and other activities. The early copy editing effort might pay off if the editorial problems were so severe that readers cannot understand the document. However, I wonder if the document is truly ready to exit the WG at such a stage.
My advice: we already pay the RFC Editor for editorial work. Let them do it. Authors may fix the editorial problems they have, particularly if they are issuing a new version. However, I think we all would be better off if reviews focused on content, and AD/author time was focused on that as well. Is something important -- such as congestion control -- missing? ABNF compiles? Behaviour rules are complete and consistent? Are the health warnings about the implications of this specification in place? And so on. I do see a lot of people typing up even editorial review comments. It may or may not be useful; something like a spelling mistake will be caught by the RFC Editor. And ADs should make absolutely sure that they really are raising blocking issues only for truly technical matters.
Jari