TODO:

* I am just going to ignore the untreated group for conducting inference
  * it is difficult to operationalize in code
  * It doesn't matter for ACRT(d) at all; for ATT(d), it doesn't matter if we interpret estimates as being nonparametric.
* Report ACRT^o and ATT^o in the summary
* change text for summary of ACRT event study



* simulations don't give expected results, not sure what is wrong
* problems with tests