feat: Rewrite Amazon scraper to use Browserless service

- Created amazon_scraper_v2.py using browserless_service (Playwright)
- Removed dependency on Crawl4AI which was not loading pages correctly
- Use browserless proxy rotation (use_proxy=True)
- More robust selector fallbacks for title, price, rating
- Fixed 2128 bytes issue - now loads full Amazon pages (>100KB)
- Updated router to use new scraper

Previous issue: Crawl4AI only loaded 2128 bytes
Now: Browserless loads complete pages with all products

app/routers/amazon.py

@@ -1,6 +1,7 @@
 """
 API Router pour la recherche Amazon France
 Utilise Server-Sent Events (SSE) pour le streaming des résultats
+Uses Browserless service for reliable scraping
 """
 
 import json
@@ -11,7 +12,7 @@ from fastapi import APIRouter, Query
 from fastapi.responses import StreamingResponse
 from pydantic import BaseModel
 
-from app.services.amazon_scraper import scrape_amazon_search, AmazonProduct
+from app.services.amazon_scraper_v2 import scrape_amazon_search, AmazonProduct
 
 logger = logging.getLogger(__name__)