Many-valued Argumentation, Conditionals and a Probabilistic Semantics for Gradual Argumentation