c should compare resampling hashes when combining benchmark results or resample results

[sebffischer]

What's wrong with the code below? Well, the resampling is not instantiated.
I think we should throw a warning when a user does this. This can be achieved by comparing the resampling hash (for resamplings instantiated on the same task).

library(mlr3verse)
#> Loading required package: mlr3

l1 = lrn("classif.ranger")
l2 = lrn("classif.rpart")

task = tsk("iris")

resampling = rsmp("holdout")
rr1 = resample(task, l1, resampling)
#> INFO  [06:59:59.544] [mlr3] Applying learner 'classif.ranger' on task 'iris' (iter 1/1)
rr2 = resample(task, l2, resampling)
#> INFO  [07:00:00.024] [mlr3] Applying learner 'classif.rpart' on task 'iris' (iter 1/1)

bmr = c(rr1, rr2)

^{Created on 2025-01-31 with reprex v2.1.1}