fix(ocr-pipeline): increase merge distance to 6% for better column merging

Sub-alignments within a column (indented words, etc.) were 60-90px apart
and not getting merged at 3%. On a typical 5-col page (~1500px), 6% = ~90px
merges sub-alignments while keeping real column boundaries (~300px) separate.

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

klausur-service/backend/cv_vocab_pipeline.py

@@ -1010,8 +1010,11 @@ def detect_column_geometry(ocr_img: np.ndarray, dewarped_bgr: np.ndarray) -> Opt
         logger.info("ColumnGeometry: < 3 clusters after verticality filter, signaling fallback")
         return None
 
-    # --- Merge clusters that are very close (3% of content width) ---
-    merge_distance = max(20, int(content_w * 0.03))
+    # --- Merge clusters that are very close ---
+    # 6% of content width: on a typical 5-col vocab page (~1500px wide),
+    # this is ~90px, which merges sub-alignments within a single column
+    # while keeping real column boundaries (~300px apart) separate.
+    merge_distance = max(30, int(content_w * 0.06))
     merged = [significant[0].copy()]
     for s in significant[1:]:
         if s['mean_x'] - merged[-1]['mean_x'] < merge_distance: