Formal Proofs as Structured Explanations: Proposing Several Tasks on Explainable Natural Language Inference