Strict script-purity tests for name_zh and name_ru

Replaces the earlier no-Latin/no-Cyrillic checks with comprehensive tests
that flag any letter character outside the target script (catches Greek,
Arabic, etc., not just Latin/Cyrillic).

Data fix: SKU name_ru contained Greek omicron (U+03BF) where Cyrillic o
(U+043E) was intended ('Скирοс' → 'Скирос').

test/test_invariants.py

@@ -358,18 +358,32 @@ def test_name_zh_no_latin(dataset):
     assert bad == [], f"name_zh with Latin letters: {bad[:10]}"
 
 
-def test_name_zh_no_cyrillic(dataset):
-    """name_zh must not contain Cyrillic characters."""
-    bad = [(e['code'], e['name_zh']) for e in dataset
-           if e.get('name_zh') and _has_cyrillic(e['name_zh'])]
-    assert bad == [], f"name_zh with Cyrillic: {bad[:10]}"
+def test_name_zh_only_cjk_letters(dataset):
+    """All letter characters in name_zh must be CJK (no Latin, Cyrillic, etc.)."""
+    bad = []
+    for e in dataset:
+        v = e.get('name_zh')
+        if not v:
+            continue
+        foreign = [c for c in v if c.isalpha() and not (
+            '㐀' <= c <= '鿿' or '豈' <= c <= '﫿'
+        )]
+        if foreign:
+            bad.append((e['code'], v, foreign))
+    assert bad == [], f"name_zh with non-CJK letters: {bad[:10]}"
 
 
-def test_name_ru_no_foreign_letters(dataset):
-    """name_ru must not contain Latin or CJK characters."""
-    bad = [(e['code'], e['name_ru']) for e in dataset
-           if e.get('name_ru') and re.search(r'[A-Za-z]', e['name_ru'])]
-    assert bad == [], f"name_ru with Latin letters: {bad[:10]}"
+def test_name_ru_only_cyrillic_letters(dataset):
+    """All letter characters in name_ru must be Cyrillic."""
+    bad = []
+    for e in dataset:
+        v = e.get('name_ru')
+        if not v:
+            continue
+        foreign = [c for c in v if c.isalpha() and not ('Ѐ' <= c <= 'ӿ')]
+        if foreign:
+            bad.append((e['code'], v, foreign))
+    assert bad == [], f"name_ru with non-Cyrillic letters: {bad[:10]}"
 
 
 def test_no_double_spaces_in_names(dataset):