fix: PDF dropdown visible + verify spelling context + araspell audit - Fix summary export dropdown clipping (overflow: visible + opens upward) - Verify: use_contextual=True in araspell_service.py (line 83) ✓ - Verify: ContextualCorrector fully initialized and used in pipeline ✓ - Verify: All araspell_rules.py classes used: AraSpellPostProcessor, ErrorClassifier, RulesBasedCorrector, OutputValidator, VocabularyManager, WordAligner, SplitMergeSpecialist, EditDistanceCorrector, ContextualCorrector, ArabicSpellChecker ✓ - Fix misleading comment about contextual=False

src/css/components.css

@@ -991,6 +991,13 @@
   background: var(--color-surface);
   border: 1px solid var(--color-summary-accent-border);
   box-shadow: var(--shadow-card);
+  overflow: visible;
+}
+
+/* Summary export dropdown opens upward */
+.summary-card .doc-dropdown__menu {
+  top: auto;
+  bottom: calc(100% + 6px);
 }
 
 .summary-card__title {
@@ -1030,6 +1037,7 @@
 
 .summary-preview.show {
   max-height: 500px;
+  overflow: visible;
 }
 
 /* ── Mobile bottom sheet ── */

src/nlp/spelling/araspell_service.py

@@ -77,7 +77,7 @@ def get_spelling_model():
         epoch = checkpoint.get('epoch', 'N/A')
         logger.info(f"Spelling model loaded on {device}, epoch: {epoch}")
 
-        # 6. Initialize the spell checker pipeline (contextual=False to save RAM)
+        # 6. Initialize the spell checker pipeline (contextual=True for MLM-based refinement)
         from nlp.spelling.araspell_rules import ArabicSpellChecker
         _spell_checker = ArabicSpellChecker(
             model, tokenizer, device, use_contextual=True