On the Epistemic Feasibility of Plans in Multiagent Systems Specifications