remote slave crashes when starting a second session

[Sylvain Pineau]

The fix for https://bugs.launchpad.net/checkbox-ng/+bug/1843326 seems to create a new issue.
Once a first session is over or cancelled, starting afresh using the same master and slave results in the following traceback on slave side:

client connection terminated abruptly
Traceback (most recent call last):
  File "/home/sylvain/lp/checkbox-ng/plainbox/vendor/rpyc/utils/server.py", line 180, in _authenticate_and_serve_client
    self._serve_client(sock2, credentials)
  File "/home/sylvain/lp/checkbox-ng/plainbox/vendor/rpyc/utils/server.py", line 201, in _serve_client
    conn = self.service._connect(Channel(SocketStream(sock)), config)
  File "/home/sylvain/lp/checkbox-ng/plainbox/vendor/rpyc/core/service.py", line 105, in _connect
    self.on_connect(conn)
  File "/home/sylvain/lp/checkbox-ng/checkbox_ng/launcher/slave.py", line 60, in on_connect
    if SessionAssistantSlave.master_blaster:
  File "/home/sylvain/lp/checkbox-ng/plainbox/vendor/rpyc/core/netref.py", line 274, in method
    return syncreq(_self, consts.HANDLE_CALLATTR, name, args, kwargs)
  File "/home/sylvain/lp/checkbox-ng/plainbox/vendor/rpyc/core/netref.py", line 76, in syncreq
    return conn.sync_request(handler, proxy, *args)
  File "/home/sylvain/lp/checkbox-ng/plainbox/vendor/rpyc/core/protocol.py", line 469, in sync_request
    return self.async_request(handler, *args, timeout=timeout).value
  File "/home/sylvain/lp/checkbox-ng/plainbox/vendor/rpyc/core/protocol.py", line 493, in async_request
    self._async_request(handler, args, res)
  File "/home/sylvain/lp/checkbox-ng/plainbox/vendor/rpyc/core/protocol.py", line 475, in _async_request
    self._send(consts.MSG_REQUEST, seq, (handler, self._box(args)))
  File "/home/sylvain/lp/checkbox-ng/plainbox/vendor/rpyc/core/protocol.py", line 259, in _send
    self._channel.send(data)
  File "/home/sylvain/lp/checkbox-ng/plainbox/vendor/rpyc/core/channel.py", line 78, in send
    self.stream.write(header + data + self.FLUSHER)
  File "/home/sylvain/lp/checkbox-ng/plainbox/vendor/rpyc/core/stream.py", line 268, in write
    count = self.sock.send(data[:self.MAX_IO_CHUNK])
  File "/home/sylvain/lp/checkbox-ng/plainbox/vendor/rpyc/core/stream.py", line 94, in __getattr__
    raise EOFError("stream has been closed")
EOFError: stream has been closed

Master just loops with:

$ checkbox-cli master 127.0.0.1 ./mylauncher
Connection lost!
connection closed by peer
Connection lost!
connection closed by peer
Connection lost!
connection closed by peer
Connection lost!
connection closed by peer

[Maciej Kisielewski (kissiel)]

Fix proposed here: https://code.launchpad.net/~kissiel/checkbox-ng/+git/checkbox-ng/+merge/382987