The value here comes from the database being able to service the websocket itself, if I read this right. So you don't have to do this in your application.
I see. I would be curious to see how that actually works out in a live environment at scale.
Not so much whether the code will be performant enough to service concurrent websocket requests, but other operational concerns such as, websocket balancing across the cluster, authentication and access logs, failure modes, or usage isolation, just to name a few things off the top of my head.