Could get_rsplit() add a method for resample_results from tune::fit_resamples()?

[mikemahoney218]

The problem

In answering tidymodels/hardhat#240 , I was tempted to use the rsample::get_rsplit() helper function on the output from tune::fit_resamples() to avoid needing to use the $splits[[1]] syntax, which I think can be a bit confusing to folks not used to thinking about rset objects. However, the outputs from fit_resamples() are of classes 'resample_results', 'tune_results', 'tbl_df', 'tbl', 'data.frame', and get_rsplit() only handles objects of class rset(). Could get_rsplit() add a method for resample_results (or, alternatively, could the output from fit_resamples() also be an rset() object)?

Reproducible example

data("ames", package = "modeldata")
ames_sf <- sf::st_as_sf(ames, coords = c("Longitude", "Latitude"), crs = 4326)

recipe <- recipes::recipe(Sale_Price ~ Year_Built, data = sf::st_drop_geometry(ames_sf)) |> 
  recipes::step_log(recipes::all_outcomes())

workflows::workflow(recipe, parsnip::linear_reg()) |> 
  tune::fit_resamples(spatialsample::spatial_clustering_cv(ames_sf)) |> 
  rsample::get_rsplit(1)
#> Error in `rsample::get_rsplit()`:
#> ! No `get_rsplit()` method for this class(es) 'resample_results', 'tune_results', 'tbl_df', 'tbl', 'data.frame'
#> Backtrace:
#>     ▆
#>  1. ├─rsample::get_rsplit(...)
#>  2. └─rsample:::get_rsplit.default(...)
#>  3.   └─rlang::abort(...)

^{Created on 2023-05-22 with reprex v2.0.2}

[hfrick]

I've moved this to tune because we usually put an S3 method for a class in the same package that owns the class. I wouldn't mind a get_rsplit.resample_results() but I'll defer to @topepo for the final answer 😄

[jrosell]

One approach could be to first extract the resamples from the tune results like I suggested here #947 and the use rsample::get_rsplit on the rset generated object..