feat: Full-Row OCR mit Spacing fuer Box-Sub-Sessions

Sub-Sessions ueberspringen Spaltenerkennung und nutzen stattdessen eine
Pseudo-Spalte ueber die volle Breite. Text wird mit proportionalem
Spacing aus Wort-Positionen rekonstruiert, um raeumliches Layout zu erhalten.

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

klausur-service/backend/ocr_pipeline_api.py

@@ -1209,6 +1209,34 @@ async def detect_columns(session_id: str):
     if img_bgr is None:
         raise HTTPException(status_code=400, detail="Crop or dewarp must be completed before column detection")
 
+    # Sub-sessions: skip column detection, create single pseudo-column
+    session = await get_session_db(session_id)
+    if session and session.get("parent_session_id"):
+        h, w = img_bgr.shape[:2]
+        column_result = {
+            "columns": [{
+                "type": "column_text",
+                "x": 0, "y": 0,
+                "width": w, "height": h,
+            }],
+            "zones": None,
+            "boxes_detected": 0,
+            "duration_seconds": 0,
+            "method": "sub_session_pseudo_column",
+        }
+        await update_session_db(
+            session_id,
+            column_result=column_result,
+            row_result=None,
+            word_result=None,
+            current_step=6,
+        )
+        cached["column_result"] = column_result
+        cached.pop("row_result", None)
+        cached.pop("word_result", None)
+        logger.info(f"OCR Pipeline: sub-session {session_id}: pseudo-column {w}x{h}px")
+        return {"session_id": session_id, **column_result}
+
     t0 = time.time()
 
     # Binarized image for layout analysis