It could happen in rare occasions that the subprocess survived death of
the tests, mostly when exceptions happened in the parent process and it
didn't kill it. So there's a finally now and the subprocess has a
limited life length just in case it still slipped somehow without kill.

Conflicts:
	src/bin/resolver/resolver.cc

Conflicts:
	doc/Doxyfile

The port used in the test was the same as the default port for the
bind10 server (5300), so the test failed if bind10 was already running on the
system. Changed it to 5321.

As suggested in the review in ticket #487, an additional non-Doxygen
comment was added to the Category enum to indicate that for error()
to work, non-error codes should come before error codes.

The tests only received data, but did not check it is correct. Now it is
checked.

Using boolean parameter instead of cryptic flag-like one. It is less
flexible, but the flexibility isn't really needed.

Removed some unused debug code and cleaned up some white space.

Pointed out in review.
A test that would catch it needs to be written.

going to push without review.

 - did it for non .cc/.h files
 - some files had 'already substitued' $Id$'s (which were probably not
   under keyword substitution with svn; just copied from some other
   file).  they were overlooked in the previous cleanups
 - there were lower cased $Id$, which were overlooked, too.

briefly talked about it on jabber, no one opposed to it, Jeremy explicitly
agreed.  changes are trivial themselves, so no explicit review.

The code analyses the response for errors as well as determining whether
it is an answer or a referral.

about this point.

still printing textual data on failure.

 - separated the zone data and find() logic
 - cleanup unnecessary variables as a result
 - used zone data for test input when possible

using the new rrsetsCheck() framework.

examining DNS messages/RRsets.