Better handling of concurrent calls to observers

We now call wg.Add(n) before we loop over n observers.  We also run the
loop inside a goroutine and therefore more quickly handle all (pid,
offset) pairs.

GetRuntimeAddresses/symbolyze/symbolyze.go

@@ -122,7 +122,8 @@ func (S *Scanner) Run() error {
 
 	proc.Close()
 
-	var wg sync.WaitGroup
+	var wg sync.WaitGroup // To be able to wait for all the observers to finish
+
 	for _, pinfo := range infos {
 		var pid_s = pinfo.Name()
 
@@ -134,9 +135,11 @@ func (S *Scanner) Run() error {
 		} else if offset, found := S.searchSymbolIn(pid); !found {
 			continue
 		} else {
-			// Notify the observers.
+			// Call the observers with (pid, offset), in the
+			// background.
+			wg.Add(len(S.observers)) // Wait for this many goroutines
+			go func() {
 				for n, observer := range S.observers {
-				wg.Add(1)
 					go func() {
 						err = observer(pid, offset)
 						if err != nil {
@@ -146,10 +149,11 @@ func (S *Scanner) Run() error {
 						wg.Done()
 					}()
 				}
+			}()
 		}
 	}
 
-	wg.Wait()
+	wg.Wait() // Wait for all observers to finish
 	return S.err
 }