Discussion thread for implementation of Bayesian model-based clustering driven by mclust().

Tuning parameters:

• Number of models/clusters
• Types of models
• Prior distribution(s)
• Tons of control parameters!

Other considerations:

• The options here could expand quickly. Perhaps we begin by focusing on Gaussian mixtures only? In fact, maybe we consider different mixture assumptions to be separate specifications?