New Windows issue (with current code including fix #48)

Hi tshirtman,

A simple stop_all() on a server is producing an error message on Win10 (not on Linux):

```
Traceback (most recent call last):
File "D:\blender-2.80-windows64\2.80\python\lib\threading.py", line 917, in _bootstrap_inner
self.run()
File "D:\blender-2.80-windows64\2.80\python\lib\threading.py", line 865, in run
self._target(*self._args, **self._kwargs)
File "C:\users\hebus\Application Data\Blender Foundation\Blender\2.80\scripts\addons\MOM\oscpy\server.py", line 336, in _listen
data, sender = sender_socket.recvfrom(65535)
OSError: [WinError 10038] Windows Error 0x2736
```
This happens despite the recent try/catch because its an OSError and not a ConnectionResetError that the new code can intercept.
I edited manually the line 337 to catch all errors as a workaround and my code can now stop the server and reconnect successfully. You will of course determinate a proper solution.

I found this while googling on the error message:
https://stackoverflow.com/questions/35889267/an-operation-was-attempted-on-something-that-is-not-a-socket-tried-fixing-a-lot

Start in server.py at the _listen method and the recvfrom call around line 336, then review the existing exception handling added for stop_all(). Reproduce the Windows stop_all() failure if possible and verify that stopping the server no longer emits the WinError 10038 traceback and that reconnecting still works.

Written by the indexing model from the issue text.