bidirectional sc methods

[kpaulfriedman]

I think that we probably need to add in a method for sc1 or sc2 that allows for the absolute value of the max_med to exceed the absolute value of the cutoff for sc endpoints that are bidirectionally interpretable. Consider testing on CEETOX endpoints potentially since these have corollary mc aeids that could be used to map to.

[brown-jason]

is there ever a situation where bmad is negative? i'm wondering if we can get away with the maximum median absolute value.