fix p2p max_peers_connected_works test

[leviathanbeak]

We've been having certain issues with this test especially in the CI,
here's the analysis and the fix.

First problem is the assertion put on the swarm itself instead of our PeerManager:

if p2p_service.swarm.connected_peers().count() > peer_limit { panic!() }

While generally this should not return more peers than allowed, one of the reasons we have PeerManager is to actually contain number of peers connected logic, so moving the assertion to

if p2p_service.peer_manager().total_peers_connected() > peer_limit { panic!() }

should suffice since PeerManager is the one disconnecting surplus peers.

Other, and bigger, issue with the test was the amount of simulated nodes we used in our example.
Having 100+ nodes running makes the test itself unnecessarily slow.
This became even more evident now that we made our Heartbeat protocol ask for a disconnect from a node that did not reply within 2 seconds with a Block Height. So nodes started disconnecting from each other and the test would never hit its exit point.

Actually there is no need to have that amount of nodes, instead same assertion can be achieved with a smaller amount, so I have decreased the amount of nodes in the test.