fix: Sync News API models with actual MongoDB schema
## 🔧 Model Synchronization
Updated Pydantic models to match actual article structure in MongoDB
### Changes
- **Article Model**: Complete restructure to match MongoDB documents
- Added Subtopic, Reference, Entities nested models
- Changed created_at to Union[str, datetime] with serializer
- Added all pipeline metadata fields (job_id, keyword_id, etc.)
- Added translation & image fields
- Changed category (single) to categories (array)
- **ArticleSummary Model**: Updated for list responses
- Synced with actual MongoDB structure
- Added news_id, categories array, images array
- **ArticleService**: Fixed category filtering
- Changed "category" to "categories" (array field)
- Updated search to include subtopics and source_keyword
- Implemented MongoDB aggregation for category list
### Verified Fields
✅ news_id, title, summary, created_at, language
✅ subtopics (array of {title, content[]})
✅ categories (array), entities (nested object)
✅ references (array), source_keyword, source_count
✅ pipeline_stages, job_id, keyword_id, processing_time
✅ images (array), image_prompt, translated_languages
### Testing
- Validated with actual English articles (20,966 total)
- Search functionality working (15,298 AI-related articles)
- Categories endpoint returning 1000+ unique categories
- All datetime fields properly serialized to ISO format
🤖 Generated with [Claude Code](https://claude.ai/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
68cc70118f