Quantifying the Expected Utility of Information in Multi-agent Scheduling Tasks