p2p/simulations: fix deadlock during network shutdown

`TestDiscoveryPersistenceSimulationSimAdapter` often got into deadlock.
The execution was stuck on two locks, i.e, `Kademlia.lock` and
`p2p/simulations.Network.lock`. Usually the test got stuck once in each
20 executions with high confidence.

`Kademlia` was stuck in `Kademlia.EachAddr()` and `Network` in
`Network.Stop()`.

Solution: in `Network.Stop()` `net.lock` must be released before
calling `node.Stop()` as stopping a node (somehow - I did not find
the exact code path) causes `Network.InitConn()` to be called from
`Kademlia.SuggestPeer()` and that blocks on `net.lock`.

Related ticket: https://github.com/ethersphere/go-ethereum/issues/1223

p2p/simulations/network.go

@@ -253,18 +253,36 @@ func (net *Network) watchPeerEvents(id enode.ID, events chan *p2p.PeerEvent, sub
 
 // Stop stops the node with the given ID
 func (net *Network) Stop(id enode.ID) error {
+	// IMPORTANT: node.Stop() must NOT be called under net.lock as
+	// node.Reachable() closure has a reference to the network and
+	// calls net.InitConn() what also locks the network. => DEADLOCK
+	// That holds until the following ticket is not resolved:
+
+	var err error
+
+	node, err := func() (*Node, error) {
 		net.lock.Lock()
 		defer net.lock.Unlock()
+
 		node := net.getNode(id)
 		if node == nil {
-		return fmt.Errorf("node %v does not exist", id)
+			return nil, fmt.Errorf("node %v does not exist", id)
 		}
 		if !node.Up() {
-		return fmt.Errorf("node %v already down", id)
+			return nil, fmt.Errorf("node %v already down", id)
 		}
 		node.SetUp(false)
+		return node, nil
+	}()
+	if err != nil {
+		return err
+	}
+
+	err = node.Stop() // must be called without net.lock
+
+	net.lock.Lock()
+	defer net.lock.Unlock()
 
-	err := node.Stop()
 	if err != nil {
 		node.SetUp(true)
 		return err