feat(kad): remove deprecated Config::set_connection_idle_timeout

[leonzchang]

Description

Related: #3844.
Related: #4656.

Notes & open questions

some tests need to be fixed, after KeepAlive::Until being removed.

Change checklist

• I have performed a self-review of my own code
• I have made corresponding changes to the documentation
• I have added tests that prove my fix is effective or that my feature works
• A changelog entry has been made in the appropriate crates

[thomaseizinger]

some tests need to be fixed, after KeepAlive::Until being removed.

To fix these tests, we need to change the following lines of code:

rust-libp2p/protocols/kad/src/behaviour.rs

Lines 1911 to 1929 in c91ecbc

	// Peer's first connection.
	if other_established == 0 {
	// Queue events for sending pending RPCs to the connected peer.
	// There can be only one pending RPC for a particular peer and query per definition.
	for (peer_id, event) in self.queries.iter_mut().filter_map(|q| {
	q.inner
	.pending_rpcs
	.iter()
	.position(|(p, _)| p == &peer_id)
	.map(|p| q.inner.pending_rpcs.remove(p))
	}) {
	self.queued_events.push_back(ToSwarm::NotifyHandler {
	peer_id,
	event,
	handler: NotifyHandler::Any,
	});
	}
	self.connected_peers.insert(peer_id);

What happens here is that after the connection is established, we tell the NetworkBehaviour about it. Without KeepAlive::Until however, the connection shuts down immediately after being established so these events are "too late". To avoid this, we need to "preload" the connection handler in the handle_established_{in,out}bound_connection callbacks. I am doing the same thing for libp2p-request-response in #4029.

This refactoring is backwards compatible so we can do that together with the deprecation of PR for set_connection_idle_timeout.

Let me know if that makes sense or you need more info :)

[leonzchang]

some tests need to be fixed, after KeepAlive::Until being removed.

Let me know if that makes sense or you need more info :)

I try to add preload_new_handler for Behaviour, but tests seems not going well, could you give me some hints?

[thomaseizinger]

Found two things so far, see if those fix the tests :)

[leonzchang]

Found two things so far, see if those fix the tests :)

Big Thanks, I found that we have to set idle_timeout when initializing swarm in tests and example.

[mergify]

This pull request has merge conflicts. Could you please resolve them @leonzchang? 🙏

[thomaseizinger]

Great! I left some inline comments.

[thomaseizinger]

Looks good, thank you!

[mergify]

This pull request has merge conflicts. Could you please resolve them @leonzchang? 🙏

[thomaseizinger]

Nice work, thank you!

[mergify]

This pull request has merge conflicts. Could you please resolve them @leonzchang? 🙏