Placement of the eval_time argument

[hfrick]

This PR moves the eval_time argument right after metrics (or metric) for user-facing functions:

• fit_resamples(), tune_grid(), tune_bayes()
• last_fit()
• fit_best()
• show_best(), select_by_pct_loss()
• autoplot()

Corresponding changes should happen in finetune for tune_race_anova(), tune_race_win_loss(), and tune_sim_anneal().

This PR does not touch .estimate_metrics() because the eval_time argument there should be removed, see #856. It also does not touch the argument order of the internal loop functions because the inconsistent naming of other arguements paired with passing arguments by position makes it likely to introduce bugs, so leaving this to #742.

The revdeps for these changes:

> revdepcheck::cloud_summary()
ℹ Syncing results to revdep/cloud.noindex/e7d0b88d-96f8-42a4-8b7e-a5ae07e69610
ℹ Comparing results
✔ agua 0.1.3                             ── E: 0     | W: 0     | N: 1    
✔ bonsai 0.2.1                           ── E: 0     | W: 0     | N: 0    
✔ finetune 1.1.0                         ── E: 1     | W: 0     | N: 0    
✔ finnts 0.4.0                           ── E: 0     | W: 0     | N: 0    
✔ healthyR.ai 0.0.13                     ── E: 1     | W: 1     | N: 0    
✔ healthyR.ts 0.3.0                      ── E: 0     | W: 0     | N: 1    
✔ HTRX 1.2.4                             ── E: 0     | W: 0     | N: 1    
✔ modeltime 1.2.8                        ── E: 0     | W: 0     | N: 1    
✔ modeltime.ensemble 1.0.3               ── E: 0     | W: 0     | N: 1    
✔ modeltime.resample 0.2.3               ── E: 0     | W: 0     | N: 1    
✔ nestedmodels 1.1.0                     ── E: 0     | W: 0     | N: 0    
✔ offsetreg 1.0.0                        ── E: 0     | W: 0     | N: 0    
✔ probably 1.0.3                         ── E: 0     | W: 0     | N: 0    
✔ shinymodels 0.1.1                      ── E: 0     | W: 0     | N: 0    
✔ stacks 1.0.3                           ── E: 0     | W: 0     | N: 0    
✔ tabnet 0.5.0                           ── E: 0     | W: 0     | N: 0    
✔ text 1.2.0                             ── E: 0     | W: 0     | N: 1    
✖ tidyclust 0.2.0                        ── E: 0  +1 | W: 0     | N: 0    
✔ tidydann 1.0.0                         ── E: 0     | W: 0     | N: 0    
✔ tidymodels 1.1.1                       ── E: 0     | W: 0     | N: 0    
✔ tidyposterior 1.0.1                    ── E: 1     | W: 0     | N: 0    
✖ tidysdm 0.9.3                          ── E: 0  +2 | W: 0  +1 | N: 0    
✔ timetk 2.9.0                           ── E: 0     | W: 0     | N: 1    
✔ usemodels 0.2.0                        ── E: 0     | W: 0     | N: 0    
✔ viralmodels 1.2.0                      ── E: 0     | W: 0     | N: 0    
✔ viruslearner 0.0.1                     ── E: 0     | W: 0     | N: 0    
✔ workflowsets 1.0.1                     ── E: 0     | W: 0     | N: 0   

[simonpcouch]

On board for mostly not touching grid_code_paths.R. Does check_initial() need a change here?

[simonpcouch]

Thanks for adding the names here!

[hfrick]

Safety! 😆

[hfrick]

Good shout! Added in ff1af86

[hfrick]

The changes for finetune are in tidymodels/finetune#104

The tests in extratests are being run for both PRs in tidymodels/extratests#201

Without the changes to check_initial() they already ran successfully. The current run includes those changes + the finetune changes.

[simonpcouch]

Looking good. :)

From my perspective, this PR and it's finetune friend are ready to merge!

I'd say we wait to file revdep breakages until we've merged the coming dots changes and possible brier score introduction in #859, as there may be some overlap.

[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.