fix(ocr-pipeline): pass left_x/right_x to classify_column_types in API path

The ocr_pipeline_api.py code path called classify_column_types without
left_x/right_x, so margin regions were never created. Also add logging
to _build_margin_regions for debugging.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

klausur-service/backend/cv_vocab_pipeline.py

@@ -2062,6 +2062,10 @@ def _build_margin_regions(
             classification_method='content_bounds',
         ))
 
+    if margins:
+        logger.info(f"Margins: {[(m.type, m.x, m.width) for m in margins]} "
+                     f"(left_x={left_x}, right_x={right_x}, img_w={img_w})")
+
     return margins
 
 

klausur-service/backend/ocr_pipeline_api.py

@@ -699,7 +699,8 @@ async def detect_columns(session_id: str):
         cached["_content_bounds"] = (left_x, right_x, top_y, bottom_y)
 
         # Phase B: Content-based classification
-        regions = classify_column_types(geometries, content_w, top_y, w, h, bottom_y)
+        regions = classify_column_types(geometries, content_w, top_y, w, h, bottom_y,
+                                        left_x=left_x, right_x=right_x)
 
     duration = time.time() - t0