avoid a false-positive assert from rebinding module

added 6 commits

• 04ce94f1...9d42f93d - 4 commits from branch master
• 5c0d514f - lib/resolve answer_finalize: separate a special case
• 6617c162 - lib/resolve answer_finalize: simplify a bit of code

Compare with previous version

We seem to be getting slightly worse results for answerrrsigs for iteration. I'm not sure whether it's significant enough to explore why is it happening.

• https://gitlab.labs.nic.cz/knot/knot-resolver/-/jobs/286594/artifacts/file/6617c162-gl1563793011_vs_master_shortlist.iter.tls6.j384.png
• https://gitlab.labs.nic.cz/knot/knot-resolver/-/jobs/286593/artifacts/file/6617c162-gl1563793010_vs_master_shortlist.iter.udp6.j384.png

It might also be good to mention/document the change of behavior when creating wire-format packet directly from lua.

Other than that, code looks good to me.

added 1 commit

• bc94b583 - kr_request: remove ::additional

Compare with previous version

added 5 commits

• bc94b583...e7899d94 - 2 commits from branch master
• 7263be7c - lib/resolve answer_finalize: separate a special case
• aaef1eca - lib/resolve answer_finalize: simplify a bit of code
• b902adaa - kr_request: remove ::additional

Compare with previous version

The answerrsigs were apparently caused by one hoster – for a subset of names, some of their NSs were supplying RRSIGs and some were not (the names do not have a trust chain in any case). I think this state has disappeared by now.

merged

mentioned in commit 9c23237b

mentioned in issue #490 (closed)

mentioned in issue #537