New method plot_trace() has been added to buc.py. This allows the user to see the distribution and the sampling evolution of the posterior parameters.
Autoregressive indexes for level, trend, and periodic-lag seasonal are now handled/recorded in _model_setup() method (i.e., no longer in the state_transition_matrix() method)
New method posterior_dict() added. It returns a dictionary that maps each posterior parameter name to its corresponding posterior samples. This is to streamline access to the posterior parameters. E.g., it's used in summary() and plot_trace() methods.
summary() method was rewritten to take advantage of posterior_dict() method.
Fixed bug where NaN values in the response array would break the model setup with a regression component. The initial starting Gibbs values for the regression coefficients now account for NaN values in the response array by ignoring the rows/indexes with NaNs.
Update to README.md to show trace plot functionality
…ly refactored, bug fixes