restore tidy.glmnet(penalty) default

[simonpcouch]

227ab92 resulted in a few new failures on extratests:

══ Failed ══════════════════════════════════════════════════════════════════════
── 1. Error ('test-encodings-glmnet.R:18:3'): parsnip models with formula interf
Error in `dimnames(x) <- dn`: length of 'dimnames' [2] not equal to array extent
Backtrace:
    ▆
 1. ├─generics::tidy(parsnip_form_fit) at test-encodings-glmnet.R:18:3
 2. └─parsnip:::tidy._elnet(parsnip_form_fit)
 3.   └─parsnip:::tidy_glmnet(x, penalty) at parsnip/R/tidy_glmnet.R:14:3
 4.     └─parsnip:::get_glmn_coefs(x$fit, penalty = penalty) at parsnip/R/tidy_glmnet.R:62:3
 5.       └─base::`colnames<-`(`*tmp*`, value = "estimate") at parsnip/R/tidy_glmnet.R:46:3

The code used to have a default penalty based on that used in the model fit; restoring that functionality.

[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.