Motivation

This pr implements peers revalidation loop.

Description

In the discovery protocol, every 10min (this could be any number actually), we need to ping back those peers whose last ping was 24hs ago to re-validate the endpoint proof, if they don't ping back, we replace them with one from the replacements list.

Changes:

• Discovery main function now spawns two tokio tasks: discover_peers_server (the one we already had) and peers_revalidation.
• UdpSocket and KademliaTable were modified to be thread/concurrent safe.
• peers_revalidation implements a tokio interval that lasts 10min and works in the following way:
  1. First we get the peers to revalidate: those that haven't pinged for 24hs.
  2. We check if each peer has been already pinged in the last iteration. If not, jump to step 3, otherwise to step 4.
  3. We invalidate the endpoint proof and ping it so that we can revalidate the proof. Finally, we pushed the peer to a list so that we can validate it in the next iteration.
  4. We check if the peer is proven. If not, it means it hasn't ping back, so we replace it with the latest pushed peer from the replacement list. If there isn't a replacement, we simply remove the peer from the bucket.

This implementation is supported by some custom unit tests and e2e tests.

Advances on issue #154.