add MAP /MLE functions

Would be good to provide users the opportunity to fit MAP() or MLE to datasets using HSSM -- based on PyMC's findMAP(). This will allow for fast debugging and cheap estimation if they don't want to be fully bayesian (or to compare).

Start by reviewing HSSM's existing fitting interface and PyMC's findMAP() behavior. Define how MAP and MLE fitting should accept datasets and how results should support debugging, inexpensive estimation, and comparison with fully Bayesian fitting; the issue names no files or tests, so the expected API and validation still need clarification.

Written by the indexing model from the issue text.