Alternative addresses are not taken into account when restarting the MCBP session

When session get socket error, it notifies bucket instance and to restart itself. It gives the its hostname, but the bucket does not take into account that the external network might be used, and might consider that the node has been ejected from the cluster config, and cancel restart process, expecting that new node will be connected upon future configuration update.

To fix the issue we should consider network name and the port to check if the node still part of the config. In addition it should pass the port number, as in cloud environment, the same address might be serving several service instances on different ports.