p2p: new dialer, peer management without locks

The most visible change is event-based dialing, which should be an
improvement over the timer-based system that we have at the moment.
The dialer gets a chance to compute new tasks whenever peers change or
dials complete. This is better than checking peers on a timer because
dials happen faster. The dialer can now make more precise decisions
about whom to dial based on the peer set and we can test those
decisions without actually opening any sockets.

Peer management is easier to test because the tests can inject
connections at checkpoints (after enc handshake, after protocol
handshake).

Most of the handshake stuff is now part of the RLPx code. It could be
exported or move to its own package because it is no longer entangled
with Server logic.

p2p/peer_test.go

@@ -28,24 +28,20 @@ var discard = Protocol{
 }
 
 func testPeer(protos []Protocol) (func(), *conn, *Peer, <-chan DiscReason) {
-	fd1, _ := net.Pipe()
-	hs1 := &protoHandshake{ID: randomID(), Version: baseProtocolVersion}
-	hs2 := &protoHandshake{ID: randomID(), Version: baseProtocolVersion}
+	fd1, fd2 := net.Pipe()
+	c1 := &conn{fd: fd1, transport: newTestTransport(randomID(), fd1)}
+	c2 := &conn{fd: fd2, transport: newTestTransport(randomID(), fd2)}
 	for _, p := range protos {
-		hs1.Caps = append(hs1.Caps, p.cap())
-		hs2.Caps = append(hs2.Caps, p.cap())
+		c1.caps = append(c1.caps, p.cap())
+		c2.caps = append(c2.caps, p.cap())
 	}
 
-	p1, p2 := MsgPipe()
-	peer := newPeer(fd1, &conn{p1, hs1}, protos)
+	peer := newPeer(c1, protos)
 	errc := make(chan DiscReason, 1)
 	go func() { errc <- peer.run() }()
 
-	closer := func() {
-		p1.Close()
-		fd1.Close()
-	}
-	return closer, &conn{p2, hs2}, peer, errc
+	closer := func() { c2.close(errors.New("close func called")) }
+	return closer, c2, peer, errc
 }
 
 func TestPeerProtoReadMsg(t *testing.T) {