Refactor choose_metric

[topepo]

Uses new code from #768. This only changes the metric code; evaluation times will be the next PR.

Generally, the error messages are improved:

old: (error) "Please check the value of metric.
new (warn): "'WAT' was not in the metric set. Please choose from: 'rmse', 'rsq'"

old (error) "Please specify a single character value for metric."
new (warn) "2 metrics were given; 'rmse' will be used"

This PR also gets rid of the old checks from the second version of tune: "The maximize argument is no longer needed."

Updated PR sequence:

1. Tools for selecting a default evaluation time #768
2. (you are here) functions to validate the chosen metric with updates to show_best() and select_best().
3. functions to validate the chosen evaluation time with updates to show_best() and select_best() (see When we need to default to a single value for eval_time #766).
4. update finetune to use the change
5. add additional functions that return (potentially) multiple evaluation times (eg autoplot())
6. refactor metric input checking code in the tune_* functions, fit_resamples(). etc.
7. write an article on tidymodels.org about this.

[hfrick]

Looks good! I've added some suggestions on error handling and cli styling since we're already at it.

[github-actions]

This pull request has been automatically locked. If you believe you have found a related problem, please file a new issue (with a reprex: https://reprex.tidyverse.org) and link to this issue.