Audit extra deck on recipes for `forested`

[hfrick]

Amongst other things, this note from the first pass at updating things to use forested

          We should look into whether this transformation is actually helpful and choose another variable / step if not.

Originally posted by @simonpcouch in #305 (comment)

[simonpcouch]

Notes from spending some time with this:

• The slides are generally quite generic and wouldn't need a major refactor.
• The slide on dates and exercise What other transformations could we do with the raw time variable? aren't relevant unless we read in a "raw" version of the dataset and convert it to dates.
• The step_unknown() step adds two zero-variance columns that are then removed by step_zv().