Reliable restart support for signaling server

[DhirajBhakta]

Issue

As of this writing, signaling server is heavily stateful, maintaining registry of active connections. Any kind of restart on this server will imply total loss of information/connections. DB is not an option as the "information" here is a websocket connection. Since the main websocket does not yet have a handler for onclose event to take care of abrupt server restarts, we lose the central registry contents

Why do we need the signaling server?

• Donee needs to get a listing of 'available servers' from a central place.
• available donor servers need to send 'heartbeat' or metrics at regular intervals to a central place.
• We need this central place to perform google Oauth login from the command line vt login

Proposed Solution

• setup the onClose handler for main socket
• continuously ping the signaling server for its "Health" API. ..do this until you get a "healthy" response.
• upon receiving healthy, re-establish the main websocket connection.

Declarative Invariants

• At all times, the central server maintains a registry of active connections to all donors/available donor servers.
• As long as the donor server is active, it will keep sending metrics to the central server