feat: peer management for protocols (with disconnection management)

[danisharora099]

This is a feature request

Problem

For Filter and LightPush, every time a new LP request/new subscription is created, getPeers() is called -- this is called every time which gives us a new set of peers to use for each request/subscription.

Further, we wish to discard a peer if it fails, and renew it with another peer. This is not trivial to do if we rely on getting a new set of peers each time.

Proposed Solutions

• Maintain a list of peers for Filter and LightPush that are updated, instead of relying on getPeers for each request: feat(lightpush): peer management for protocols #2003

LightPush:

• Renew a peer if fails to send a message: feat(lightpush): peer management for protocols #2003

Filter:

• Use peer management: feat(filter): use protocol peer management #2047
• Renew peer if filter ping fails on a node: feat(filter): peer/subscription renewal with recurring Filter pings #2052
• Check if messages are missing from one node but not another, replace a peer that is missing to deliver messages: feat: validate messages for individual filter nodes & perform renewals #2057

Notes

• part of [Epic: js-waku] Reliability Protocol for Resource-Restricted Clients #2154

[chaitanyaprem]

Renew peer if filter ping fails on a node
Not sure if we can depend on this as expected filter ping duration is 1 minute.
Maybe better to listen on disconnection signal from a peer and then act on it, which can be based on periodic ping that is run more often. In go-waku/status-go all connected peers are pinged every 10 seconds.

[weboko]

marking as done as all PR are complete