fix: address open triage issues (#21,#26,#45,#46,#47,#30-33)

- #45: re-read local folder when recursive checkbox toggles
- #47: clear Identify crop spinner when image is already cached
- #46: search selected people by person_ids; AND for "First Last"
- #21: enqueue network import without blocking API walk
- #26: serve resized grid thumbnails for photos
- #30/#31/#33: re-enable small-face filters in auto-match

backend/api/photos.py

@@ -37,7 +37,6 @@ from backend.schemas.search import (
     SearchPhotosResponse,
 )
 from backend.services.photo_service import (
-    find_photos_in_folder,
     import_photo_from_path,
 )
 from backend.services.search_service import (
@@ -63,6 +62,10 @@ def search_photos(
     current_user: Annotated[dict, Depends(get_current_user)],
     search_type: str = Query("name", description="Search type: name, date, tags, no_faces, no_tags, processed, unprocessed, favorites"),
     person_name: Optional[str] = Query(None, description="Person name for name search"),
+    person_ids: Optional[str] = Query(
+        None,
+        description="Comma-separated person IDs for exact name search (preferred over substring when selecting people)",
+    ),
     tag_names: Optional[str] = Query(None, description="Comma-separated tag names for tag search"),
     match_all: bool = Query(False, description="Match all tags (for tag search)"),
     date_from: Optional[str] = Query(None, description="Date from (YYYY-MM-DD)"),
@@ -111,13 +114,36 @@ def search_photos(
         tag_list = [t.strip() for t in tag_names.split(",") if t.strip()]
     
     if search_type == "name":
-        if not person_name:
+        parsed_person_ids: list[int] = []
+        if person_ids:
+            for part in person_ids.split(","):
+                part = part.strip()
+                if not part:
+                    continue
+                try:
+                    parsed_person_ids.append(int(part))
+                except ValueError:
+                    raise HTTPException(
+                        status_code=status.HTTP_400_BAD_REQUEST,
+                        detail=f"Invalid person_ids value: {part}",
+                    )
+        if not person_name and not parsed_person_ids:
             raise HTTPException(
                 status_code=status.HTTP_400_BAD_REQUEST,
-                detail="person_name is required for name search",
+                detail="person_name or person_ids is required for name search",
             )
         results, total = search_photos_by_name(
-            db, person_name, folder_path, media_type, df, dt, tag_list, match_all, page, page_size
+            db,
+            person_name or "",
+            folder_path,
+            media_type,
+            df,
+            dt,
+            tag_list,
+            match_all,
+            page,
+            page_size,
+            person_ids=parsed_person_ids or None,
         )
         for photo, full_name in results:
             tags = get_photo_tags(db, photo.id)
@@ -360,23 +386,20 @@ def import_photos(
             detail=f"Folder not found: {request.folder_path}",
         )
 
-    # Estimate number of photos (quick scan)
-    estimated_photos = len(find_photos_in_folder(request.folder_path, request.recursive))
-
-    # Enqueue job
-    # Pass function as string path to avoid serialization issues
+    # Do NOT walk the tree on the API thread (#21) — that blocked SharePoint/SMB
+    # imports for minutes with no progress. Counting happens inside the RQ job.
     job = queue.enqueue(
         "backend.services.tasks.import_photos_task",
         request.folder_path,
         request.recursive,
-        job_timeout="1h",  # Allow up to 1 hour for large imports
+        job_timeout="2h",  # Network shares can be slow
     )
 
     return PhotoImportResponse(
         job_id=job.id,
         message=f"Photo import job queued for {request.folder_path}",
         folder_path=request.folder_path,
-        estimated_photos=estimated_photos,
+        estimated_photos=None,
     )