Enable GPU by default for PaddleOCR

- Changed use_gpu default from False to True
- Added use_gpu parameter to PaddleOCR init
- Added show_log=False to reduce log noise

GPU acceleration significantly improves OCR performance and
reduces memory pressure when processing scanned PDFs.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

src/ocr/paddle_ocr.py

@@ -45,7 +45,7 @@ class OCREngine:
     def __init__(
         self,
         lang: str = "en",
-        use_gpu: bool = False,
+        use_gpu: bool = True,  # Default to GPU for better performance
         det_model_dir: str | None = None,
         rec_model_dir: str | None = None
     ):
@@ -54,20 +54,25 @@ class OCREngine:
 
         Args:
             lang: Language code ('en', 'sv', 'ch', etc.)
-            use_gpu: Whether to use GPU acceleration
+            use_gpu: Whether to use GPU acceleration (default: True)
             det_model_dir: Custom detection model directory
             rec_model_dir: Custom recognition model directory
         """
         from paddleocr import PaddleOCR
 
-        # PaddleOCR 3.x API - simplified init
-        init_params = {'lang': lang}
+        # PaddleOCR init with GPU support
+        init_params = {
+            'lang': lang,
+            'use_gpu': use_gpu,
+            'show_log': False,  # Reduce log noise
+        }
         if det_model_dir:
             init_params['text_detection_model_dir'] = det_model_dir
         if rec_model_dir:
             init_params['text_recognition_model_dir'] = rec_model_dir
 
         self.ocr = PaddleOCR(**init_params)
+        self.use_gpu = use_gpu
 
     def extract_from_image(
         self,