Fix OSError in NLTK check and use HistGradientBoostingClassifier

utils/preprocessing.py

@@ -11,7 +11,7 @@ from typing import List, Optional
 # Download required NLTK data (handled gracefully for HF Spaces)
 try:
     nltk.data.find("tokenizers/punkt_tab")
-except LookupError:
+except (LookupError, OSError):
     nltk.download("punkt_tab", quiet=True)
 try:
     nltk.data.find("corpora/stopwords")