bug: excess relay connections are not being properly pruned

[gabrielmer]

Problem

In the waku.sandbox fleet, we noticed that relay peers were consuming all the configured libp2p connections.

Looking further at the logs, we found that there's an excess of in connections not being properly taken care of

    
topics="waku node peer_manager" tid=1 file=peer_manager.nim:981 inRelayConns=252/134 outRelayConns=46/66 totalConnections=300/300 notConnectedPeers=2028 outsideBackoffPeers=2028

Expected behavior

The node should properly prune excess peer connections and respect its targets

[gabrielmer]

Reopening the issue as now that we controlled in connections, we see excess out connections in waku.sandbox
Investigating it