Was ignoring RRSIGs related to a RRSet that was completely removed.
This was a relict of old code when RRSIGs were connected to their
RRSets.

Also simplified the function
(and function sign_rr_should_be_signed()) and did some minor
refactoring and polishing.

Zscanner Building

DNSSEC fixes

Two fixes in this MR:

1. Continuation of fix merged in !176. Knot was still generating RRSIGs for the old (wrong) NSEC3s, as it did not know that they would be removed. Fixed by marking nodes from which NSEC3s are to be removed, so that these are not signed in the next step. (Commit 38aeecc3)
2. Function for saving removed NSEC/NSEC3 records into changeset (`knot_nsec_changeset_remove()`) was ignoring NSEC3 RRSIGs. (Commit 8b5f08ac). This probably manifested after UPDATE - a test should be added for this case.

Log output of runtests

Hattrie Leq Fix

pure hashtable is a node type where all keys share a common letter,
so it is consumed on parent node -> pure trie node transition.
So when tracing back, this letter must be returned.

Also updated unit tests to cover these sort of situations.

Systemd

Was not storing RRSIGs of NSEC3s.

Was leaving RRSIGs of removed NSEC3s in the zone, if some NSEC3
changed.

issue #213

issue #213

close #213

issue #213

Main Cleanup

- reorganized
- some code blocks are split
- SIGALRM handler for main thread

Cov #1187492, #1187493, #1187494, #1187495, #1138581, #1138582

Nsec No Prove Wildcards Below Dp

Buildsystem Fixes

- libtap is now built separately
- libtap is not linked into server
- tests are built separately
- no need to maintain list of tests in a file