Return posterior distribution from cdf function

[cdeweyx]

For multi models, the user can pass ci=True and cdf() returns a tuple with the mean, lower bound, and upper bound of the confidence interval.

In use cases where one wants to compare across groups, returning the posterior distribution itself is more desirable than summary metrics.

I'd like to add a parameter for the user to specify that they want the posterior distributions for each group. I'll link to the PR below.

[erikbern]

that makes sense conceptually, but we're just sampling from the posterior, so it will just be a crude estimate of the cdf. but i'm fine having a flag so it returns the full set of samples rather than a summary

[cdeweyx]

Deprioritizing this for now!