server: asynchronously handle internal broadcast/send requests

This commit modifies the request handling within the sever’s
queryHandler goroutine to ensure that requests from the ChannelRouter
or other related sub-systems don’t block the main processing loop.

We do this simply by launching a goroutine to handle the dispatch of
the request.

server.go

@@ -578,6 +578,11 @@ out:
 
 			srvrLog.Debugf("Broadcasting %v messages", len(bMsg.msgs))
 
+			// Launch a new goroutine to handle the broadcast
+			// request, this allows us process this request
+			// asynchronously without blocking subsequent broadcast
+			// requests.
+			go func() {
 				s.peersMtx.RLock()
 				for _, peer := range s.peersByPub {
 					if ignore != nil &&
@@ -596,6 +601,7 @@ out:
 				s.peersMtx.RUnlock()
 
 				bMsg.errChan <- nil
+			}()
 		case sMsg := <-s.sendRequests:
 			// TODO(roasbeef): use [33]byte everywhere instead
 			//  * eliminate usage of mutexes, funnel all peer
@@ -605,6 +611,10 @@ out:
 			srvrLog.Debugf("Attempting to send msgs %v to: %x",
 				len(sMsg.msgs), target)
 
+			// Launch a new goroutine to handle this send request,
+			// this allows us process this request asynchronously
+			// without blocking future send requests.
+			go func() {
 				s.peersMtx.RLock()
 				targetPeer, ok := s.peersByPub[string(target)]
 				if !ok {
@@ -612,7 +622,7 @@ out:
 					srvrLog.Errorf("unable to send message to %x, "+
 						"peer not found", target)
 					sMsg.errChan <- errors.New("peer not found")
-				continue
+					return
 				}
 
 				for _, msg := range sMsg.msgs {
@@ -621,6 +631,7 @@ out:
 				s.peersMtx.RUnlock()
 
 				sMsg.errChan <- nil
+			}()
 		case query := <-s.queries:
 			switch msg := query.(type) {
 			case *connectPeerMsg:
@@ -695,6 +706,7 @@ func (s *server) handleConnectPeer(msg *connectPeerMsg) {
 			Addr:      addr,
 			Permanent: true,
 		})
+		msg.err <- nil
 	} else {
 		// If we're not making a persistent connection, then we'll
 		// attempt to connect o the target peer, returning an error