Using a randomized controlled trial to test whether modifications to contingency management improve outcomes for heavy drinkers with serious mental illness