Use dict.copy() for thread safety, better atomicity speed and memory efficiency

We adapted a bunch of code to be really thread-safe, espacially when iterating over a dict . It also improve atomicity speed and the code is more memory efficient now. For instance, that code is not completely thread-safe: for item in list ( self .my_queue.values()): # ... And such code is not immunte to "dictionary changed size during iteration" errors. Here is the new code: for item in self .my_queue.copy().values(): # ... it applies to all operations on dict objects: dict.items() , dict.keys() and dict.values() .