Add payment line parser and fix OCR override from payment_line

- Add MachineCodeParser for Swedish invoice payment line parsing
- Fix OCR Reference extraction by normalizing account number spaces
- Add cross-validation tests for pipeline and field_extractor
- Update UI layout for compact upload and full-width results

Key changes:
- machine_code_parser.py: Handle spaces in Bankgiro numbers (e.g. "78 2 1 713")
- pipeline.py: OCR and Amount override from payment_line, BG/PG comparison only
- field_extractor.py: Improved invoice number normalization
- app.py: Responsive UI layout changes

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

src/yolo/annotation_generator.py

@@ -22,6 +22,7 @@ FIELD_CLASSES = {
     'Amount': 6,
     'supplier_organisation_number': 7,
     'customer_number': 8,
+    'payment_line': 9,  # Machine code payment line at bottom of invoice
 }
 
 # Fields that need matching but map to other YOLO classes
@@ -43,6 +44,7 @@ CLASS_NAMES = [
     'amount',
     'supplier_org_number',
     'customer_number',
+    'payment_line',  # Machine code payment line at bottom of invoice
 ]
 
 
@@ -160,6 +162,68 @@ class AnnotationGenerator:
 
         return annotations
 
+    def add_payment_line_annotation(
+        self,
+        annotations: list[YOLOAnnotation],
+        payment_line_bbox: tuple[float, float, float, float],
+        confidence: float,
+        image_width: float,
+        image_height: float,
+        dpi: int = 300
+    ) -> list[YOLOAnnotation]:
+        """
+        Add payment_line annotation from machine code parser result.
+
+        Args:
+            annotations: Existing list of annotations to append to
+            payment_line_bbox: Bounding box (x0, y0, x1, y1) in PDF coordinates
+            confidence: Confidence score from machine code parser
+            image_width: Width of the rendered image in pixels
+            image_height: Height of the rendered image in pixels
+            dpi: DPI used for rendering
+
+        Returns:
+            Updated annotations list with payment_line annotation added
+        """
+        if not payment_line_bbox or confidence < self.min_confidence:
+            return annotations
+
+        # Scale factor to convert PDF points (72 DPI) to rendered pixels
+        scale = dpi / 72.0
+
+        x0, y0, x1, y1 = payment_line_bbox
+        x0, y0, x1, y1 = x0 * scale, y0 * scale, x1 * scale, y1 * scale
+
+        # Add absolute padding
+        pad = self.bbox_padding_px
+        x0 = max(0, x0 - pad)
+        y0 = max(0, y0 - pad)
+        x1 = min(image_width, x1 + pad)
+        y1 = min(image_height, y1 + pad)
+
+        # Convert to YOLO format (normalized center + size)
+        x_center = (x0 + x1) / 2 / image_width
+        y_center = (y0 + y1) / 2 / image_height
+        width = (x1 - x0) / image_width
+        height = (y1 - y0) / image_height
+
+        # Clamp values to 0-1
+        x_center = max(0, min(1, x_center))
+        y_center = max(0, min(1, y_center))
+        width = max(0, min(1, width))
+        height = max(0, min(1, height))
+
+        annotations.append(YOLOAnnotation(
+            class_id=FIELD_CLASSES['payment_line'],
+            x_center=x_center,
+            y_center=y_center,
+            width=width,
+            height=height,
+            confidence=confidence
+        ))
+
+        return annotations
+
     def save_annotations(
         self,
         annotations: list[YOLOAnnotation],