S3 registration clean-up

[hfrick]

closes #872

Most of the functions registered in .onLoad() were already registered in the regular way, receipts below. The autoplot() methods for model_fit and glmnet were not but since parsnip imports ggplot2, this PR switches this to "regular registration" as well.

Already registered in the regular way

• tidy.model_fit()
• tidy.nullmodel()
• tidy._elnet()
• tidy._lognet()
• tidy._multnet()
• tidy._fishnet()
• glance.model_fit()
• augment.model_fit()
• required_pkgs.model_fit()
• required_pkgs.model_spec()

[simonpcouch]

This looks great! Thanks for linking to those existing tags, saved me a good amount of time in review.

For posterity:

vctrs::s3_register() docs seem to indicate this is a good move for us. I also poked around blames here for a bit, and

• For the tidier methods, it seems like this pattern first existed only because we previously registered the s3 methods against broom's generic, which was neither Imported nor Suggested at the time.
• As for autoplot, this commit justified the usage of s3_register() with "glmnet is not a formal dependency"—while the autoplot method is for the class that glmnet defines, we don't actually use any glmnet machinery in its internals.

Goodbye, unneeded complexity!💭

[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.