My idea is that it could be built on top if the relay communication is flexible enough. most nostr libraries build the auth, limitations, and connection policies at the "new WebSocket()" level, but that makes it difficult to add additional types of behavior that a different client might want later

For example you could implement a relay blacklist at the websocket level so that it would throw an error (or make it look offline) if the client connected to a blacklisted relay. or you could extend that relay pool class to ensure it never even tries to connect to blacklisted relays. all of this is just a theory though, I still have to put it into practice to see if it works