API output without Persian diacritics or written ezafe ی (company decision); tutor answers too

plain_persian() strips kasra/fatha/damma/shadda/sukun/hamza-above and the written «ه‌ی» ezafe, keeping tanwin. Applied
at output, not in data: PlainPersianJSONRenderer runs one byte-level pass over every JSON response (~5 ms per MB
locally), and to_plain() cleans tutor answers including the stream. Rewriting titles in the DB/seed would have broken
question binding by title path and SeedRecord keys. Tests updated to expect the cleaned text; config/tests_plain_persian.py.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

rag/tests_tutor_quality.py

@@ -209,7 +209,7 @@ class ConversationMemoryTests(TestCase):
         self.assertIn('توان منفی', retrieve_mock.call_args.args[0])
 
     def test_a_follow_up_is_never_served_from_someone_elses_cache(self):
-        replies = iter(['جوابِ اولِ الف', 'جوابِ «نفهمیدم»ِ الف', 'جوابِ اولِ ب', 'جوابِ «نفهمیدم»ِ ب'])
+        replies = iter(['جوابِ اولِ الف', 'جوابِ «نفهمیدم»ِ الف', 'جوابِ اولِ ب', 'جواب «نفهمیدم» ب'])
         retrieve = patch('rag.services.retrieval.retrieve_context',
                          return_value=([], {'gradeTitle': 'نهم'}))
         generate = patch('rag.services.generation.generate_answer',
@@ -221,7 +221,7 @@ class ConversationMemoryTests(TestCase):
             conv_b = self._ask(b, 'تجزیه چیست؟').json()['conversationId']
             second_b = self._ask(b, 'نفهمیدم', conv_b)
 
-        self.assertEqual(second_b.json()['answer'], 'جوابِ «نفهمیدم»ِ ب')
+        self.assertEqual(second_b.json()['answer'], 'جواب «نفهمیدم» ب')
 
 
 @override_settings(**RAG_ON)