Sensitivity analysis for average treatment effects in matched observational studies