Assessing the robustness of time-to-event models for estimating unmarked wildlife abundance using remote cameras