More thoughts on chapter orders

[malcolmbarrett]

Ideas:

• ~~Move sensitivity to after outcome model, then cover specialty sensitivity analyses in the relevant chapters~~: I no longer like this idea because I don't think this will work out neatly
• Collapse the two exposure chapters, then g-comp, then doubly robust without ML
• Move the mediation chapter before survival because it flows from the time-varying example.
• But maybe mediation BEFORE time-varying because TV can be an expansion of a three-timepoint example in mediation