feat: add GPU optimization and fix TableData consistency

GPU Optimization (Section 3.1):
- Add comprehensive memory management for RTX 4060 8GB
- Enable all recognition features (chart, formula, table, seal, text)
- Implement model cache with auto-unload for idle models
- Add memory monitoring and warning system

Bug Fix (Section 3.3):
- Fix TableData field inconsistency: 'columns' -> 'cols'
- Remove invalid 'html' and 'extracted_text' parameters
- Add proper TableCell conversion in _convert_table_data

Documentation:
- Add Future Improvements section for batch processing enhancement

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

backend/app/core/config.py

@@ -63,10 +63,41 @@ class Settings(BaseSettings):
         return [lang.strip() for lang in self.ocr_languages.split(",")]
 
     # ===== GPU Acceleration Configuration =====
+    # Basic GPU settings
     force_cpu_mode: bool = Field(default=False)
-    gpu_memory_fraction: float = Field(default=0.8)
+    gpu_memory_fraction: float = Field(default=0.7)  # Optimized for RTX 4060 8GB
     gpu_device_id: int = Field(default=0)
 
+    # Memory management for RTX 4060 8GB
+    gpu_memory_limit_mb: int = Field(default=6144)  # 6GB max for models (leave 2GB buffer)
+    gpu_memory_reserve_mb: int = Field(default=512)  # Reserve for CUDA overhead
+    enable_memory_optimization: bool = Field(default=True)
+
+    # Model loading and caching
+    enable_lazy_model_loading: bool = Field(default=True)  # Load models on demand
+    enable_model_cache: bool = Field(default=True)
+    model_cache_limit_mb: int = Field(default=4096)  # Max 4GB for cached models
+    auto_unload_unused_models: bool = Field(default=True)  # Unload unused language models
+    model_idle_timeout_seconds: int = Field(default=300)  # Unload after 5 min idle
+
+    # Batch processing configuration
+    enable_batch_processing: bool = Field(default=True)
+    inference_batch_size: int = Field(default=1)  # Conservative for 8GB VRAM
+    max_concurrent_pages: int = Field(default=2)  # Process 2 pages concurrently
+
+    # PP-StructureV3 optimization
+    enable_chart_recognition: bool = Field(default=True)  # Chart/diagram recognition
+    enable_formula_recognition: bool = Field(default=True)  # Math formula recognition
+    enable_table_recognition: bool = Field(default=True)  # Table structure recognition
+    enable_seal_recognition: bool = Field(default=True)  # Seal/stamp recognition
+    enable_text_recognition: bool = Field(default=True)  # General text recognition
+    layout_detection_threshold: float = Field(default=0.5)
+
+    # Performance tuning
+    use_fp16_inference: bool = Field(default=False)  # Half-precision (if supported)
+    enable_cudnn_benchmark: bool = Field(default=True)  # Optimize convolution algorithms
+    num_threads: int = Field(default=4)  # CPU threads for preprocessing
+
     # ===== File Upload Configuration =====
     max_upload_size: int = Field(default=52428800)  # 50MB
     allowed_extensions: str = Field(default="png,jpg,jpeg,pdf,bmp,tiff,doc,docx,ppt,pptx")