Intended usage: Quick resolution attempt with an empty cache.

export QNAME=...
export QTYPE=...
sudo -E docker run -e QNAME -e QTYPE krestest:latest '-n' '-c' '/etc/knot-resolver/kresd.conf'

It creates new callback functions for every request which
uses "callback chaining" but these should be rare.

It seems there is no reason to keep this function private in policy
module.

It did not log through request->trace_log and is not very useful anyway.

DEBUG_IF accepts user-supplied function which decides which requests
should be logged.

Attempt to avoid duplicating ten lines in debug_logfinish_cb lead me
to splitting kr_log_qverbose_impl into two functions kr_log_q and kr_log_req.
This is another minor change to API exposed to modules.

Formerly both logs used slightly different formats and duplicated code.
From now on verbose log and request tracing are generated using the same
code.

This required a small change to request trace_log_f definition so it
might affect external modules.

We de-randomize qname letter case very early so kr_pkt_text() should see
all lowercase. Any uppercase means mismatch.

I've considered moving debug print to an earlier place but this way it
is more visible so I like more.

fix strict aliasing problems

See merge request !962

I tested that older compilers don't refuse this option:
 - gcc 4.8 (as that's the default in CentOS 7)
 - llvm 3.8 (year 2016 is enough, I hope)

The issue here is that `char *` is not allowed to alias
with `anyType *`.  With gcc-10 in Fedora this now started
to cause real problems and loading stats module segfaulted.

Actually I can't see in standard (C11 6.5 par.7) that using `void *`
is guaranteed to be correct, but at least it seems fine with gcc,
and e.g. some standard functions like posix_memalign() use it
in the same "dangerous" way.

ci/debian-buster: Add docker-py to docker image

See merge request !961

ci/debian-buster: add libnghttp2 to docker image

See merge request !960

distro/tests: failure to refresh Leap15 repo is non-critical

See merge request !959

doc: fix a broken internal link

See merge request !958

Reported on Turris forums:
https://forum.turris.cz/t/dns-over-tcp-just-a-single-transaction/12003/21

Thanks to Tomas.

compiler warning improvements

See merge request !952

ci: disable build check for EPEL8 (still unresolvable)

See merge request !953

Normally the types are equal, but that's not true e.g.
on 32-bit systems and macOS.

I just saw a case with bad luajit includes that linked successfully even
with some of the macros getting this warning.