fix: handle jikan rate limits gracefully instead of dropping data

internal/features/anime/service.go

@@ -79,8 +79,9 @@ func (s *Service) GetWatchingAnime(ctx context.Context, userID string) ([]templa
 	for _, row := range rows {
 		anime, err := s.jikanClient.GetAnimeByID(int(row.AnimeID))
 		if err != nil {
-			// Skip if we can't fetch anime details
-			continue
+			// Instead of skipping, we still append it, but without the extra Jikan details
+			// This prevents anime from vanishing from the watchlist when Jikan rate limits us.
+			anime = jikan.Anime{}
 		}
 		result = append(result, templates.WatchingAnimeWithDetails{
 			Entry: row,

internal/worker/relations.go

@@ -71,7 +71,16 @@ func (w *Worker) syncRelations(ctx context.Context) {
 
 	for _, a := range animes {
 		func() {
-			// Always mark as synced and sleep so the queue advances even on error.
+			animeData, err := w.client.GetAnimeByID(int(a.ID))
+			if err != nil {
+				log.Printf("worker: failed to fetch anime details for %d: %v", a.ID, err)
+				// Sleep a bit on error to respect rate limits, but DO NOT mark as synced
+				// so it will be retried on the next worker run.
+				time.Sleep(2 * time.Second)
+				return
+			}
+
+			// If we got here, we successfully fetched the data, so we mark it as synced.
 			defer func() {
 				err := w.db.MarkRelationsSynced(ctx, a.ID)
 				if err != nil {
@@ -80,12 +89,6 @@ func (w *Worker) syncRelations(ctx context.Context) {
 				time.Sleep(400 * time.Millisecond)
 			}()
 
-			animeData, err := w.client.GetAnimeByID(int(a.ID))
-			if err != nil {
-				log.Printf("worker: failed to fetch anime details for %d: %v", a.ID, err)
-				return
-			}
-
 			for _, rel := range animeData.Relations {
 				for _, entry := range rel.Entry {
 					if entry.Type == "anime" {