Conditional effects in the CTF

[jmafoster1]

CATEs are currently fairly basic. We should change this to pass in a filter parameter to generate a stratum of the population which can then be passed to any of the basic effect metrics.
E.g. If we want to estimate the effect of X on Y for people older than 50, we could pass in lambda age: age > 50. This would then filter the dataframe and pass this stratum of data into the effect measure (e.g. ATE, risk ratio, etc.).