Anfonwch hwn fel neges destun: On guaranteed optimal robust explanations for NLP models