Implement sampling routine to recover marginalized RVs from posterior trace

[ricardoV94]

Related to #91

@zaxtax mentions that the function for computing the logps may be equally useful, if not even more useful, to users. So we should make those easily accessible as well.

Would be nice to extend the change-point notebook to include the plots in the STAN doc: https://mc-stan.org/docs/stan-users-guide/change-point.html

[ricardoV94]

Closed via #285