Change shift_peak so you can input general distribution and not necessarily energy resolution

Currently to shift_peak the input is true and pred and it automatically calculates a resolution from that. We should change that.