Preserve grammar annotations (pl), (no pl) and skip articles in IPA

Two fixes:
1. Add pl, sg, no, also, ae, be etc. to _GRAMMAR_BRACKET_WORDS so
   annotations like (pl) and (no pl) are not replaced with IPA.
2. Skip articles (the, a, an) in fix_ipa_continuation_cell — they
   never get IPA in vocabulary books.

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

klausur-service/backend/cv_ocr_engines.py

@@ -888,6 +888,10 @@ _GRAMMAR_BRACKET_WORDS = frozenset({
     'up', 'out', 'off', 'into', 'over', 'down', 'away', 'back', 'through',
     # English grammar abbreviations used in vocab tables
     'sth', 'sb', 'adj', 'adv',
+    # Number/plural/grammar annotations
+    'pl', 'sg', 'sing', 'no', 'also', 'auch',
+    # Regional English markers
+    'ae', 'be', 'ame', 'bre',
 })
 
 
@@ -1293,10 +1297,11 @@ def fix_ipa_continuation_cell(
         return garbled_text
 
     # 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.
+    # Skip articles (the, a, an) — they never get IPA in vocab books.
+    # Other function words like "down", "up" are kept because they are
+    # integral parts of phrasal verbs (e.g. "close down").
     # Skip words that already have inline IPA in the headword row.
+    _ARTICLES = {'the', 'a', 'an'}
     ipa_parts: List[str] = []
     for part in parts:
         # A part may be multi-word like "secondary school"
@@ -1308,6 +1313,8 @@ def fix_ipa_continuation_cell(
                 continue
             if covered_words and clean_w.lower() in covered_words:
                 continue  # Already has IPA inline in the headword
+            if clean_w.lower() in _ARTICLES:
+                continue  # Articles never get IPA in vocab books
             ipa = _lookup_ipa(clean_w, pronunciation)
             if ipa:
                 word_ipas.append(ipa)

klausur-service/backend/tests/test_grid_editor_api.py

@@ -485,13 +485,14 @@ class TestGarbledIpaDetection:
         assert _text_has_garbled_ipa("equipment") is False
 
     def test_fix_continuation_united_kingdom(self):
-        """IPA continuation for 'the United Kingdom' → proper IPA."""
+        """IPA continuation for 'the United Kingdom' → IPA without 'the'."""
         fixed = fix_ipa_continuation_cell(
             "[n, nn]", "the United Kingdom", pronunciation="british",
         )
         # Should contain proper IPA, not the garbled text
         assert fixed != "[n, nn]"
         assert "kˈɪŋdəm" in fixed  # Kingdom IPA
+        assert "ðə" not in fixed    # "the" must NOT get IPA
 
     def test_fix_continuation_equipment(self):
         """IPA continuation for 'equipment' → proper IPA."""