per convention.

don't duplicate the duplicate zone check in sqlite3 accessor's addZone()
added documentation

Note: currently called createZone so as not to interfere with an existing addZone() call in MemoryClient.
Instead of pure virtual I made it have a default 'NotImplemented' method.
Also, the DatabaseClient holds the transaction (not the SQLite3Accessor); we could do it on that level but we already have a transaction API on a higher level and it makes a little bit more sense to me to do it there

this will suppress a cppcheck error:
http://git.bind10.isc.org/~tester/builder//BIND10-cppcheck/20121209143001-FreeBSD8-amd64-GCC/logs/cppcheck.out

okayed on jabber.

The test used to create template files in a testdata folder without write
permissions. That caused 'make distcheck' errors. This regression had
been introduced with the ticket #2479. Okayed on jabber.

cppcheck was complaining about what was effectively:
const int A = 6; const int B = A; std::string array[b];

Changing this to
const int B = 6; std::string array[b];

... works

This reflects the fact that at the moment, the data is stored in
memory and not in a backing file.

in docs. trivial. no review.

okayed on jabber.

with fixing Conflicts for:
	src/lib/dns/gen-rdatacode.py.in
	src/lib/dns/tests/rdata_txt_like_unittest.cc

When tree.find() results in SUBDOMAIN, the node_path's top contains the
previous node.  Calling previousNode() yet again will return its
further previous node.

This test currently fails with an assertion.