p2p/nodestate: fix deadlock during shutdown of les server (ethereum#2…

…1927)

This PR fixes a deadlock reported here: ethereum#21925

The cause is that many operations may be pending, but if the close happens, only one of them gets awoken and exits, the others remain waiting for a signal that never comes.

p2p/nodestate/nodestate.go

@@ -725,7 +725,7 @@ func (ns *NodeStateMachine) opFinish() {
 	}
 	ns.opPending = nil
 	ns.opFlag = false
-	ns.opWait.Signal()
+	ns.opWait.Broadcast()
 }
 
 // Operation calls the given function as an operation callback. This allows the caller