Fix IPA continuation: only process fully-bracketed cells, keep phrasal verb particles

Two fixes:
1. Step 5d now only treats cells as continuation when text is entirely
   inside brackets (e.g. "[n, nn]"). Cells with headwords outside brackets
   (e.g. "employee [im'ploi:]") are no longer overwritten.
2. fix_ipa_continuation_cell no longer skips grammar words like "down" —
   they are part of the headword in phrasal verbs like "close sth. down".

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

klausur-service/backend/cv_ocr_engines.py

@@ -1266,7 +1266,10 @@ def fix_ipa_continuation_cell(
     if not parts:
         return garbled_text
 
-    # Look up IPA for each headword part
+    # Look up IPA for each headword part.
+    # Do NOT skip grammar words here — they are integral parts of the
+    # headword (e.g. "close down", "the United Kingdom").  Grammar
+    # annotations like "(sth)", "(no pl)" are already stripped above.
     ipa_parts: List[str] = []
     for part in parts:
         # A part may be multi-word like "secondary school"
@@ -1276,9 +1279,6 @@ def fix_ipa_continuation_cell(
             clean_w = re.sub(r'[^a-zA-Z\'-]', '', w)
             if not clean_w or len(clean_w) < 2:
                 continue
-            # Skip grammar words like "to" at the start
-            if clean_w.lower() in _GRAMMAR_BRACKET_WORDS:
-                continue
             ipa = _lookup_ipa(clean_w, pronunciation)
             if ipa:
                 word_ipas.append(ipa)