Fix false headers on sparse layouts and IPA corruption on German text

1. Header detection: Add 25% cap to single-cell heading heuristic.
   On German synonym dicts where most rows naturally have only 1
   content cell, the old logic marked 60%+ of rows as headers.

2. IPA de/all mode: Use "column_text" (light processing) for non-
   English columns instead of "column_en" (full processing). The
   full path runs _insert_missing_ipa() which splits on whitespace,
   matches English prefixes ("bildschön" → "bild"), and truncates
   the rest — destroying German comma-separated synonym lists.

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

klausur-service/backend/grid_editor_helpers.py

@@ -913,6 +913,18 @@ def _detect_heading_rows_by_single_cell(
                 continue
             heading_row_indices.append(ri)
 
+        # Guard: if >25% of eligible rows would become headings, the
+        # heuristic is misfiring (e.g. sparse single-column layout where
+        # most rows naturally have only 1 content cell).
+        eligible_rows = len(non_header_rows) - 2  # minus first/last excluded
+        if eligible_rows > 0 and len(heading_row_indices) > eligible_rows * 0.25:
+            logger.debug(
+                "Skipping single-cell heading detection for zone %s: "
+                "%d/%d rows would be headings (>25%%)",
+                z.get("zone_index"), len(heading_row_indices), eligible_rows,
+            )
+            continue
+
         for hri in heading_row_indices:
             header_cells = [c for c in cells if c.get("row_index") == hri]
             if not header_cells: