fix: handle IsNoMatchError for remediator startWatch (#1734) (#1736)

This error can occur when the RESTMapper fails to look up the object
type, and has been observed to happen semi-frequently for the e2e
TestCRDDeleteBeforeRemoveCustomResourceV1.

The e2e test is flaky because depending on the ordering of events, the
conflict metric may not be recorded. This should fix the metric
flakiness for this edge case.

Co-authored-by: Sam Dowell <[email protected]>

Author: tiffanny29631

pkg/remediator/watch/filteredwatcher.go

@@ -344,7 +344,7 @@ func (w *filteredWatcher) start(ctx context.Context, resourceVersion string) (bo
 	if err != nil {
 		if errors.Is(err, context.Canceled) || errors.Is(err, context.DeadlineExceeded) {
 			return false, status.InternalWrapf(err, "failed to start remediator watch for %s", w.gvk)
-		} else if apierrors.IsNotFound(err) {
+		} else if apierrors.IsNotFound(err) || meta.IsNoMatchError(err) {
 			statusErr := syncerclient.ConflictWatchResourceDoesNotExist(err, w.gvk)
 			klog.Warningf("Remediator encountered a resource conflict: "+
 				"%v. To resolve the conflict, the remediator will enqueue a resync "+