Fix more logger related memory leaks on Windows

[skizunov]

Some observations have been made on Windows:

• Any process (besides the main process) that does not call
  salt.log.setup.setup_multiprocessing_logging() will leak. The reason is
  that because Windows doesn't copy process state when spawning processes,
  each new process will have the temp null and temp queue log handlers
  installed (the default state in salt/log/setup.py when initializing).
  Those handlers just store stuff and hence will use much memory.
• It was obversed that for some of the classes, register_after_fork()
  wouldn't work and so salt.log.setup.setup_multiprocessing_logging()
  wouldn't be invoked causing leaks. However, for some of the classes,
  register_after_fork() would work. After investigation, it was determined
  that register_after_fork() worked only on processes that used
  __setstate__ / __getstate__ and hence would invoke __init__() inside
  the child process. Thus, register_after_fork() only seems to work on
  Windows when invoked from within the child process, but not the parent
  process, making it not very useful (since if we are in the child process,
  we can invoke salt.log.setup.setup_multiprocessing_logging() directly).

The following changes have been made based on these observations:

• Try to make almost every process use or inherit from
  MultiprocessingProcess so as to call
  salt.log.setup.setup_multiprocessing_logging().
• Processes that don't use or inherit from MultiprocessingProcess invoke
  salt.log.setup.setup_multiprocessing_logging() themselves.
• Try to make every process that uses or inherits from
  MultiprocessingProcess invoke __init__ from the child process.
  In order to accomplish this, all such processes use __setstate__ and
  __getstate__.
• When multiprocessing.Process is constructed via __init__ in the child
  process, certain features such as is_alive() and ident() don't work.
  Thus we set self._is_child = True in each __setstate__ to be able to
  distinguish whether the __init__ invocation is from the parent or child
  process.

Signed-off-by: Sergey Kizunov sergey.kizunov@ni.com

[cachedout]

This looks totally sensible to me. Nice work, @skizunov

I would also like to run this past @s0undt3ch before we merge this.

[s0undt3ch]

This seems pretty reasonable to me and an excellent work from @skizunov.

Let's just make sure that all of salt's daemons exit properly on CTRL-C and when signalled with SIG_TERM and SIG_INT, that's my only concern.

Thank You @skizunov!