This option allows binding to the same port using separate
IPv6 and IPv4 socket.

refs #1877

refs #1877

- Freeing RRSIGs after RRSet deep copy (there's no switch that would mean no RRSIGs copyiing)
- Added assert to zone-dump that checks for such state

Refs #1752

refs #1968

refs #1969

- Added to zones section (configuration reference), might be moved later

Refs #1877 @2h

- Was causing compilation error with latest OpenSSL (v1.0.1C)
- Made rdata conversion functions static.

- Disabled where I was able to tell it's OK.
- Enabled in zone-diff, as I'm not sure it would work without it.

Should mitigate problems with IXFRs.

Conflicts:
	src/libknot/updates/xfr-in.c

- Wrong flag was used.
- Added some detailed debug messages (may be removed later).

refs #1952 @3h

- If the first RRSet was empty and the second one contained only
  one RDATA, this RDATA would get deallocated.

- Storing each RDATA separately during changeset application. The
  chains may get corrupted due to other changes and thus may not be
  deallocated properly. This caused leaks in some cases when the
  IXFR failed.
  This includes counting how many space we must reserve by counting
  number of RDATA to be stored.
- Function for getting RDATA count from the RDATA chain itself

- Returning new node or NULL (if empty) in dname_get_node().
- Modified dname_update_node() to consider empty nodes.
- Freeing empty nodes after everything is finished.
- Marking empty nodes by a new flag.

- Having RRSIG reference in SOA RRs in changeset resulted in wrong stream size calculation
- Added function for changeset dump (pretty format - same as in zone)
- Fixed wrong debug definitons

Refs #1949 @1.5h

- Added more detailed debug messages to zone-diff

- In IXFR/AXFR processing replaced rrset_merge() by
  rrset_merge_no_dupl().
- In some other cases it is not reuqired.
- Modified rrset_compare() function to not check TTL
  TODO: check if this doesn't cause problems elsewehere.
- Needs some testing.

refs #1953 @20m

- If all RDATA have been removed from a RRSet, but the RRSet was
  not removed because RRSIGs remained, and new RDATA added to this
  RRSet had different TTL, the TTL from the old RRSet was used.
  This was fixed by rewriting the TTL by the TTL fo the new RRSet.

refs @1963 @30m

+ Some minor fixes.

refs #1949