SimulPL: Aligning Human Preferences in Simultaneous Machine Translation