Don't mark node down when control connection fails to connect

Node pools should be stable, if cc fails to connect it is not good
enough reason to neither to kill it nor to mark node down.

Author: dkropachev

cassandra/cluster.py

@@ -3546,10 +3546,6 @@ def _connect_host_in_lbp(self):
         for host in lbp.make_query_plan():
             try:
                 return (self._try_connect(host), None)
-            except ConnectionException as exc:
-                errors[str(host.endpoint)] = exc
-                log.warning("[control connection] Error connecting to %s:", host, exc_info=True)
-                self._cluster.signal_connection_failure(host, exc, is_host_addition=False)
             except Exception as exc:
                 errors[str(host.endpoint)] = exc
                 log.warning("[control connection] Error connecting to %s:", host, exc_info=True)