Extend possibilities to calculate marginal likelihood

[florianhartig]

Following discussion here http://stats.stackexchange.com/questions/209810/computation-of-the-marginal-likelihood-from-mcmc-samples/210196

see also http://stats.stackexchange.com/questions/44972/what-is-the-difference-between-laplace-approximation-and-modified-harmonic-me

[florianhartig]

Also consider to include bridge sampling, see

https://cran.rstudio.com/web/packages/bridgesampling/index.html

https://github.com/quentingronau/bridgesampling/tree/stan_interface

[florianhartig]

Check also https://arxiv.org/abs/1704.03472

[TankredO]

Added bridge sampling: 155b24bcad9390a4680d6951e514da2f750385d7 e20102c6c205e4dc98cff09becd152d22c009f02 a1f858df24084a02b46bd87725bacf3cdab1b352

[florianhartig]

@TankredO, are the current options fully documented, and do we also check them all in the unit tests?

[TankredO]

All options are documented. Added simple test cases for all four methods: a4994696ffc43b523782ced4b82583f9a9465ff8