feat: add callbacks for resample and benchmark

[be-marc]

No description provided.

[be-marc]

callback = callback_workhorse(
  id = "mlr3.selected_features",
  on_workhorse_before_result = function(callback, context) {
    context$env$learner$state$selected_features = setdiff(context$env$learner$model$frame$var, "<leaf>")
  }
)

learner = lrn("classif.rpart")
task = tsk("iris")
resampling = rsmp("cv", folds = 3)

rr = resample(task, learner, resampling, callbacks = callback)

rr$learners[[1]]$state$selected_features