Planned for merge #2268

And make it slightly stronger (it now checks something that should be
mostly obvious too, but who knows, with software).

Instead of releasing it directly. While the internal release was more
convenient, it didn't allow for swapping things fast under a mutex and
then spending the time releasing it unlocked.

It is supposed to be mostly private. Use it internally only.

this will prevent surprising test failure on an incrementally updated dev
tree.  okayed on jabber.

* The unique constraint on NSEC3 owner per zone is now removed
* NSEC3 RRSIGs are now returned next to their RR set
* Test was adjusted to check for RRSIGs too

They don't need the mutable versions to work, but they used them anyway.

This one was not mutable, it just took Name as an argument.

As the method was removed.

The domain tree no longer supports the mutable find versions, so we
don't need the template parameters allowing for them.

They are currently used in test code only (so we'll need to update the
tests to compile). The real code seams clean now.

Currently, the content of zone is passed to the addZone method. No way
to change it later, but the addZone overwrites the old value if called
again.

This should hopefully allow for removal of mutable find methods from the
tree.

with the cleanup in this ticket we can now actually return NULL, but
intermediate status of other #2283 tickets will require non NULL values.
So, it's probably better to keep the behavior with referring to #2283,
instead of #2284.

Unfortunately it turned out it doesn't work for the originally intended
purpose.  See the ticket for details.

this addresses the issue described in #2291.  Still not really clean,
but thanks to boost::bind we can centralize all the code logic in the
callback, so I think it's now less likely that we forget cleaning it up
when the hack is not necessary.

as it's now completely stateless and can work independently.
the common specialization for the main implementation is defined in a
new created .cc file.