wcfs: Fix false alarm about faulty client whereas client just restarted

This patch fixes false alarm issue which tests were added in previous
patch 775adf73.

wcfs/wcfs.go

@@ -2063,8 +2063,19 @@ func (wlink *WatchLink) shutdown(reason error) {
 
 		// kill client if shutdown is due to faulty pin handling
 		if kill {
+			// Prevent misleading kill messages / statistics:
+			//
+			// Double check if client didn't stop already due to other
+			// reason and therefore doesn't reply to pin request.
+			//
+			// => give 1 second for client to terminate to prevent race
+			//    between ctx.Cancel and terminating client.
+			ctx, _ := context.WithTimeout(context.Background(), 1*time.Second)
+			_, err := waitProcessEnd(ctx, wlink.client)
+			if err {
 				wlink.badPinKill(reason) // only fatal error
 			}
+		}
 
 		// NOTE unregistering watches and wlink itself is done on serve exit, not
 		//      here, to avoid AB-BA deadlock on atMu and e.g. WatchLink.byfileMu .

wcfs/wcfs_faultyprot_test.py

@@ -480,7 +480,6 @@ def _bad_watch_killed_by_other(ctx, f, at):
     _bad_watch_suicide(ctx, f, at)
     f.assertKilled(ctx, "test didn't kill bad client")
 
-@mark.xfail
 @mark.parametrize('faulty,dokill', [
     [_bad_watch_suicide, False],
     [_bad_watch_killed_by_other, True],
@@ -539,7 +538,6 @@ def _bad_pin_killed_by_other(ctx, f, at):
     _bad_pin_suicide(ctx, f, at)
     f.assertKilled(ctx, "test did not kill")
 
-@mark.xfail
 @mark.parametrize('faulty,dokill', [
     [_bad_pin_suicide, False],
     [_bad_pin_killed_by_other, True],