A new way to try and deal with deadlocks that occur in app_queue at present....

  A new way to try and deal with deadlocks that occur in app_queue at present. Using this approach, we only manipulate the main queue mutexes when we get a dev state change on a device that is actually a member of a queue. Further optimizations are still possible (eg - store and manage pointers to the status integer of the member record that this interface/device has a one-to-one relationship with and then go directly to those pointers to make status modifications rather than the recursive looping that goes on now) BUT first things first. :)



git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@30430 65c4cc65-6c06-0410-ace0-fbb531ad65f3