Loop and wait for multiple terminted processes.

If you have a bunch of exec() calls (translated into a bunch of additional processes created), upon their termiantion OpenSIPS will get a cascade of SIGCHLD signals - as the kernel does not guarantee the delivery of the signals (like mergining similar multiple signal triggers), we will not have a 1-to-1 between terminated processes and received signals.
So, let's do our best and upon a single SIGCHLD we will try to "catch" as many terminated processes as possible/available.

Closes #1773

(cherry picked from commit 57f3619)

main.c

@@ -684,9 +684,9 @@ static void sig_usr(int signo)
 					exit(0);
 					break;
 			case SIGCHLD:
-					pid = waitpid(-1, &status, WNOHANG);
-					LM_DBG("SIGCHLD received from %ld (status=%d), ignoring\n",
-						(long)pid,status);
+					while ( (pid = waitpid(-1, &status, WNOHANG))>0 )
+						LM_DBG("SIGCHLD received from %ld (status=%d),"
+							" ignoring\n", (long)pid,status);
 					break;
 			case SIGSEGV:
 					/* looks like we ate some spicy SIP */