as an argument of get(), inc() and start_timer()

this chunk was accidentally removed in e448bbba
causing test faiulres.

because it is a message ID, add a changelog entry too.
(reviewed via jabber)

reviewed by jelte via jabber.

Changes some log message descriptions and some
function arguments and some output.

It should raises again an exception after incrementing the counter if it's
raised while initializing ThreadingUnixStreamServer.

Due to the the method name changed

Due to the class name changed

[2225_xfrout] update creating the counter object and invoking the incrementer method according to the new interface

moved the counter from handle_request() to _select_loop() and revised the
related unit test

XfroutCounter class and TestXfroutCounter are removed

Because self._counter_xfrrej is no longer assigned to `None` as a
default in the constructor.

 - used find() instead of `in` operator for exact pattern matching of string
 - a style fix in the list

 - added incrementer/decrementer for the number of IXFR/AXFR running into
   the
   XfroutCounter class

 - added getter methods of counter handlers for the XfroutSession/NotifyOut
   class into the XfroutCounter class

 - inserted incrementer/decrementer handler for the number of AXFR/IXFR
   running into XfroutSession

 - removed the unnecessary thread lock from the get_statistics()

 - removed a commented-out code

 - updated copyright

 - If it omits setting the counters, which are notifyoutv4, notifyoutv6,
   xfrrej, and xfrreqdone, they are set to None as defaults when the object is
   initiating. Then when each counter is invoked in some method, it checks
   whether the counter is Nonetype or not. Unless the counter is NoneType, it
   invokes the counter.  After that unless the counter is callable, a TypeError
   exception would be raised.

 - added some tests for this changes

 - removed some dead code from xfrout.py.in

- added the new class XfroutCounter
  It is intended for producing incrementers of the counters based on the
  statistics spec and for outputting statistics data format to passed to the
  b10-stats daemon.
- added counter handlers passed to the NotifyOut object and the XfroutSession
  object
- added handling of the getstats command
- added counting xfr requests in the XfroutSession class
- updated the UnixSockServer class for taking over the counter handlers

Normally, notify_out finds A/AAAA records for the rdata in NS records
and uses the non-primary ones as slaves. For these, it uses a hardcoded
port value of 53 which won't work in lettuce as these tests use
non-standard ports.

This is because CPython < version 3.2 does not support it.

This reverts commit 9040c490.

This reverts commit e295b40a.

it's not actually been observed, and even if it continues the loop
it's probably less harmful.  but it should make more sense to terminate
the loop and switch to the new session in this case.

I did some other cleanups:
- log the failure regardless of the exact error code returned by recv_fd
  (in fact, when it fails it should always be FD_SYSTEM_ERROR in practice).
- remove the code comment about that.  It talked about rather unusual case
  and could be misleading.  And, now the detailed version of log message
  provides sufficient information.
- revise the detailed log message so it will explain more likely reason
  for that and clarify what the admin should do.

no behavior change.