feat: auto-persist metadata and chapter list on first book preview visit

When a book is not in the local DB and the preview endpoint is hit,
metadata and the chapter index are now saved to PocketBase in the
background. Subsequent visits load from the local store instead of
scraping live. Chapter text is not fetched until an explicit scrape job
is triggered.

scraper/internal/server/handlers_preview.go

@@ -13,6 +13,7 @@ package server
 //	GET  /api/chapter-text-preview/{slug}/{n}   — scrape a single chapter text live
 
 import (
+	"context"
 	"encoding/json"
 	"fmt"
 	"net/http"
@@ -75,6 +76,26 @@ func (s *Server) handleBookPreview(w http.ResponseWriter, r *http.Request) {
 		chapters = []scraper.ChapterRef{}
 	}
 
+	// If the book was not already in the library, persist the metadata and
+	// chapter list skeleton to PocketBase now so that subsequent visits load
+	// from the local store rather than scraping live again.  Chapter text is
+	// NOT fetched here — that still requires an explicit scrape job.
+	if !inLib {
+		go func() {
+			bgCtx := context.Background()
+			if werr := s.store.WriteMetadata(bgCtx, meta); werr != nil {
+				s.log.Warn("book-preview: WriteMetadata failed (non-fatal)", "slug", slug, "err", werr)
+			}
+			if len(chapters) > 0 {
+				if werr := s.store.WriteChapterRefs(bgCtx, slug, chapters); werr != nil {
+					s.log.Warn("book-preview: WriteChapterRefs failed (non-fatal)", "slug", slug, "err", werr)
+				}
+			}
+			s.log.Info("book-preview: metadata+chapter list persisted", "slug", slug, "chapters", len(chapters))
+		}()
+		inLib = true // will be true by the time the client navigates back
+	}
+
 	resp := BookPreviewResponse{
 		InLib:    inLib,
 		Meta:     meta,