Subjects per grade (math, calculus, geometry, stats, discrete), 5 new textbooks, 202 new islands, long islands split, hand-written figures and lesson texts for the new books

- content.Subject + Unit.subject, /api/home?subject=, tutor grounded in the chosen book
- 5 new books ingested; ~1150 hand-written questions, islands named by hand
- node_split: islands over 7 lessons split into parts of <=6 (lesson ids and progress kept)
- Lesson.figures hand-authored and visually checked for 653 lessons of the new books;
  gen_lesson_figures got --subject and survives network errors
- seed files re-exported

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

rag/management/commands/rag_model_benchmark.py

@@ -125,7 +125,7 @@ class Command(BaseCommand):
         parser.add_argument('--judge', default=DEFAULT_JUDGE)
         parser.add_argument('--per-grade', type=int, default=6)
         parser.add_argument('--collections', nargs='*', default=None,
-                            help='پیش‌فرض: همه‌ی کتاب‌های ریاضی/حسابان')
+                            help='پیش‌فرض: همه‌ی کتاب‌های ریاضی/حسابان/هندسه/آمار/گسسته')
         parser.add_argument('--seed', default='bench-v1')
         parser.add_argument('--workers', type=int, default=4)
         parser.add_argument('--no-context', action='store_true',
@@ -149,7 +149,7 @@ class Command(BaseCommand):
         if collections:
             qs = qs.filter(collection__in=collections)
         else:
-            qs = qs.filter(collection__regex=r'^(ریاضی|حسابان) ')
+            qs = qs.filter(collection__regex=r'^(ریاضی|حسابان|هندسه|آمار و احتمال|گسسته) ')
         by_collection = defaultdict(list)
         for chunk in qs.only('id', 'problem', 'checked_answer', 'collection', 'grade_title',
                              'subject', 'chapter'):