put in a full outline

[davidwhogg]

I see things like

• When do you need MCMC?
• M-H, emcee, and references to competitor methods like nested and hamiltionian etc.
• using the output to make figures, get numbers with error bars, and show results
• tuning and Gibbs sampling and the like
• how do you know you have run long enough?
• initialization and burn-in
• Don'ts and no-nos and breaking Markov-ness.

[davidwhogg]

done; about to push..