Support missForest() imputation method

[joeycouse]

It would be really cool if recipes could support missing variable imputation with the missForest package

Good example here: https://rpubs.com/lmorgan95/MissForest#:~:text=MissForest%20is%20a%20random%20forest,then%20predicts%20the%20missing%20part.

On another note, I'm wondering if there would be value in a general step_impute() API with user supplied "method" e.g. mean, median, mode, knn, bag_tree, mice, linear, etc. Similar to how parsnip handles different engine arguments. Outside the scope of this issue but food for thought.

If you'd be interesting in a step_impute_missForest() function let me know and I can try to work on a PR. Thanks!

[EmilHvitfeldt]

On another note, I'm wondering if there would be value in a general step_impute() API with user supplied "method" e.g. mean, median, mode, knn, bag_tree, mice, linear, etc. Similar to how parsnip handles different engine arguments. Outside the scope of this issue but food for thought.

I already have that idea listed here: EmilHvitfeldt/extrasteps#66 😄