Bias bounds and target trials for causal inference in observational epidemiology