Allow exploration of Beta parameters through a Pert distribution

[bakaburg1]

It may be simpler from some people use a harnessing the Pert distribution to describe the shape of their priors. It would also provide a cleaner solution in the generation of the Beta parameters since there is an analytical conversion between Pert and Beta. This could also help when the optimization algorithm of get_beta_params() fails to converge.

It may be wise then to create a new function based on the Pert.