This is such a random argument, and you bring it up in every thread about PGP. No, it's not at all the case that authenticated encryption is only meaningful in "stateful, online protocols", nor is it the case that signatures and authentication tags are interchangeable. Even PGP doesn't agree with this weird argument.