Need to take a lock when accessing listeners hash table because listeners can be removed by other threads. Must use a recursive mutex because the notify calls can then come back and call gst_inter_pipe_leave_node() or gst_inter_pipe_add_node() which results in a 2nd attempt to lock the listeners mutex in the same thread.
Fix for https://github.com/RidgeRun/gst-interpipe/issues/116
Need to take a lock when accessing listeners hash table because listeners can be removed by other threads. Must use a recursive mutex because the notify calls can then come back and call gst_inter_pipe_leave_node() or gst_inter_pipe_add_node() which results in a 2nd attempt to lock the listeners mutex in the same thread.