Fixes 4.7

I am starting to make PRs of my personal branch of 4.7. This is a series of isolated individual changes. Maybe worth some discussion for 22ef0b14, where I changed the behaviour of load_mh_file with mh_replic=0. In such a case, I find it useful that posterior statistics of estimated parameters are always computed in oo_. If for some reason something went wrong in a previous estimation (for example parallel chains finished their jobs, but the master crashed) and the oo_ is not updated to the last batch MCMC iterations, there would be no other way to recover the posterior statistics of parameters from GetPosteriorParametersStatistics.m into oo_. In other words, this would cover situations where oo_ is not synchronized with the last status of MCMC. Other commits should be straightforward.