feat: refactor summarizer and PDF extraction pipeline

- Split summarizer into summary_generator and summary_persister modules
- Refactor pdf_image_extractor to two-phase pipeline with PicoDet layout detection
- Add layout_detector service for PicoDet-S_layout_3cls integration
- Add exceptions module with ConflictError and NotFoundError
- Improve admin dashboard with better statistics and task management
- Add design review document with system optimization suggestions
- Add new tests for crawler, pdf_downloader, pipeline, and summary_utils
- Update dependencies and configuration
- Clean up dead code and improve error handling

app/config.py

@@ -27,6 +27,7 @@ class Settings(BaseSettings):
     HTTP_TIMEOUT_SECONDS: int = 30
     HTTP_MAX_RETRIES: int = 3
     HTTP_USER_AGENT: str = "hf-daily-papers-local/0.1"
+    PDF_DOWNLOAD_TIMEOUT: int = 120
 
     # AI 总结
     SUMMARY_BACKEND: str = "pi"  # "pi" | "claude"
@@ -36,7 +37,9 @@ class Settings(BaseSettings):
     SUMMARY_CONCURRENCY: int = 3
     SUMMARY_TIMEOUT_SECONDS: int = 1200
     SUMMARY_MAX_RETRIES: int = 2
-    SUMMARY_PDF_MODE: str = "auto"  # "auto" = ≤80k 用 inject，>80k 用 search；也可强制 "inject" / "search"
+    SUMMARY_PDF_MODE: str = (
+        "auto"  # "auto" = ≤80k 用 inject，>80k 用 search；也可强制 "inject" / "search"
+    )
 
     # 调度
     SCHEDULER_ENABLED: bool = False
@@ -56,6 +59,10 @@ class Settings(BaseSettings):
     EMBED_MODEL: str = ""
     EMBED_DIMENSIONS: int = 0
 
+    # 布局检测
+    LAYOUT_MODEL_PATH: str = "data/models/picodet_layout_3cls.onnx"
+    LAYOUT_THRESHOLD: float = 0.5
+
     model_config = {
         "env_file": str(BASE_DIR / ".env"),
         "env_file_encoding": "utf-8",