Produce old-school caret's bwplots

[jpm92]

Hello, I'm trying to evaluate graphically a workflow_set after a tuning process and I've seen a lot of publications using caret's bwplot() function, which I think are quite nice. I was surprised I couldn't find anything similar on tidymodels.
Ideally, I would like to compare in a boxplot the metrics for each of the best set of hyperparameters from each model that has been tuned.
I have achieved the results after cherry-picking the model's metrics from the workflow_set tuning results, but this is tedious and very prone to errors.

nnet <- models_tuned %>%
  unnest(result) %>%
  unnest(.metrics) %>%
  filter(wflow_id=="norm_nnet",
         .iter==0,
         .config=="Preprocessor1_Model02",
         .metric %in% c("roc_auc", "f_meas", "sens", "spec")
  ) %>%
  select(wflow_id, id, id2, .metric, .estimate)

rfor <- models_tuned %>%
  unnest(result) %>%
  unnest(.metrics) %>%
  filter(wflow_id=="norm_rfor",
         .iter==2,
         .metric %in% c("roc_auc", "f_meas", "sens", "spec")
  ) %>%
  select(wflow_id, id, id2, .metric, .estimate)

knn <- models_tuned %>%
  unnest(result) %>%
  unnest(.metrics) %>%
  filter(wflow_id=="norm_knn",
         .iter==0,
         .config=="Preprocessor1_Model02",
         .metric %in% c("roc_auc", "f_meas", "sens", "spec")
  ) %>%
  select(wflow_id, id, id2, .metric, .estimate)

glm <- models_tuned %>%
  unnest(result) %>%
  unnest(.metrics) %>%
  filter(wflow_id=="norm_glm",
         .config=="Preprocessor1_Model01",
         .metric %in% c("roc_auc", "f_meas", "sens", "spec")
  ) %>%
  select(wflow_id, id, id2, .metric, .estimate)

svm <- models_tuned %>%
  unnest(result) %>%
  unnest(.metrics) %>%
  filter(wflow_id=="norm_svm",
         .iter==2,
         .metric %in% c("roc_auc", "f_meas", "sens", "spec")
  ) %>%
  select(wflow_id, id, id2, .metric, .estimate)

clean <- bind_rows(nnet, rfor, knn, glm, svm)

clean %>%
  group_by(wflow_id, .metric) %>%
  summarise(mean=mean(.estimate, na.rm=T), n=n())

ggplot(clean, aes(.estimate, wflow_id)) +
  geom_boxplot(na.rm=T) +
  facet_wrap(~.metric)

Is there a better way to do this? I know I can get some graphs with autoplot(), but they don't give that much information.
Thanks a lot!