####rand_int : implements the random interventional analogue to direct and indirect natural effects
####ipw_med : implements calculations of controlled direct effect with IPTW, optionally setting M to different values or distributions.
####med_reg : implements regression-based calculation of CDE, NDE, and NIE
Everything assumes
- No X -> Y unmeasured confounder
- No M -> Y unmeasured confounder
Addtionally
3. med_reg assumes no exposure-induced M -> Y confounder
4. NDE and their randomized analogue assume no X -> M confounder
###Reference. VanderWeele, T. (2015). Explanation in causal inference: methods for mediation and interaction. Oxford University Press.