benchmark should work with learner containing TuneToken if it is overwritten by param_values

[mb706]

library("mlr3")
library("paradox")
bmg <- benchmark_grid(tsk("iris"), lrn("classif.rpart", cp = to_tune()), rsmp("holdout"), param_values = list(list(list(cp = 1))))
benchmark(bmg)
#> Error: <LearnerClassifRpart:classif.rpart> cannot be trained with TuneToken present in hyperparameter: cp

This error is not necessary, since the cp hyperparameter is overwritten given the param_values:

bmg <- benchmark_grid(tsk("iris"), lrn("classif.rpart"), rsmp("holdout"), param_values = list(list(list(cp = 1))))
benchmark(bmg)$learners$learner[[1]]$param_set$values
#> INFO  [22:49:43.505] [mlr3] Running benchmark with 1 resampling iterations
#> INFO  [22:49:43.520] [mlr3] Applying learner 'classif.rpart' on task 'iris' (iter 1/1)
#> INFO  [22:49:43.546] [mlr3] Finished benchmark
#> $cp
#> [1] 1
#> 
#> $xval
#> [1] 0
#>