node: Don't panic on connection logic error

It seems like the underlying network library has a bug, and transport
handovers don't always happen.

radicle-node/src/service.rs

@@ -1335,6 +1335,12 @@ where
                     // In this scenario, it's possible that our peer is persistent, and
                     // disconnected. We get an inbound connection before we attempt a re-connection,
                     // and therefore we treat it as a regular inbound connection.
+                    //
+                    // It's also possible that a disconnection hasn't gone through yet and our
+                    // peer is still in connected state here, while a new inbound connection from
+                    // that same peer is made. This results in a new connection from a peer that is
+                    // already connected from the perspective of the service. This appears to be
+                    // a bug in the underlying networking library.
                     let peer = e.get_mut();
                     debug!(
                         target: "service",

radicle-node/src/service/session.rs

@@ -231,7 +231,7 @@ impl Session {
         self.last_active = since;
 
         if let State::Connected { .. } = &self.state {
-            panic!("Session::to_connected: session is already in 'connected' state");
+            log::error!(target: "service", "Session {} is already in 'connected' state, resetting..", self.id);
         };
         self.state = State::Connected {
             since,

radicle-node/src/wire/protocol.rs

@@ -491,7 +491,7 @@ where
                 &DisconnectReason::connection(),
             );
         } else {
-            log::warn!(target: "wire", "Tried to cleanup unknown peer with id={id} (fd={fd})");
+            log::debug!(target: "wire", "Tried to cleanup unknown peer with id={id} (fd={fd})");
         }
     }
 }