tetragon: Cleanup tgids array before another round of events iteration

We might not get all events immediately on the first events iteration,
so there needs to be another round.

In that case we need to zero tgids array and return an error, so the
checker knows we failed and starts another events iteration.

Signed-off-by: Jiri Olsa <jolsa@kernel.org>

pkg/sensors/exec/exec_test.go

@@ -212,7 +212,13 @@ func TestEventExitThreads(t *testing.T) {
 	finalCheck := func(l *logrus.Logger) error {
 		// Make sure we saw all pids
 		for pid, used := range tgids {
-			assert.True(t, used, "did not see exit event for pid %d", pid)
+			if !assert.True(t, used, "did not see exit event for pid %d", pid) {
+				// Cleanup the 'seen' tgids, because we are starting over
+				for pid := range tgids {
+					tgids[pid] = false
+				}
+				return fmt.Errorf("final check failed")
+			}
 		}
 		return nil
 	}