Add error for glmnet models if penalty is not exactly 1

[juliasilge]

Closes #481

This PR adds a new error for glmnet models, checking if the length() of penalty is exactly one in the translate() functions. Here is what I have as the error message for now:

library(tidymodels)
#> Registered S3 method overwritten by 'tune':
#>   method                   from   
#>   required_pkgs.model_spec parsnip
data("two_class_dat")


logistic_reg(penalty = c(0.01, 0.1)) %>% 
  set_engine("glmnet") %>% 
  fit(Class ~ ., data = two_class_dat)
#> Error: For the glmnet engine, `penalty` must be a single number.
#> * There are 2 values for `penalty`.
#> * To try multiple values for total regularization, use the tune package.


logistic_reg() %>% 
  set_engine("glmnet") %>% 
  fit(Class ~ ., data = two_class_dat)
#> Error: For the glmnet engine, `penalty` must be a single number.
#> * There are 0 values for `penalty`.
#> * To try multiple values for total regularization, use the tune package.

^{Created on 2021-05-06 by the reprex package (v2.0.0)}

Thoughts? I have it saying to try tune even for the no-penalty case because people are probably thinking it will just go do something automatic.

[juliasilge]

I have changed quite a lot of examples and docs now that translate() fails without a penalty for glmnet. Take a look and see if you don't like any of these choices I made.

[hfrick]

@EmilHvitfeldt pointed out that it might be little surprising to receive a message about prediction (i.e. the reference to multi_predict() when calling fit().

tidymodels/censored#55 (review)

[juliasilge]

We chatted about this and decided we are comfortable with this message for now 👍 because we expect one of the main reasons folks might try to put multiple values for penalty in is to be able to predict at different values; we want to be sure folks know how to get those results out.

[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.