feat: rFerns classif learner from old mlr2

[AnnaNzrv]

Thank you for contributing a learner to the mlr3 ecosystem.
Please make sure that:

• The added learner(s) are sufficiently tested
• All the CI tests are passing (including the CodeFactor)
• You ran devtools::document()
• You updated the NEWS.md field to include the addition of the learner
• You did not modify anything not related to the new learner
• You are listed as a contributor in the DESCRIPTION of the R package

[sebffischer]

@AnnaNzrv can you investigate why changing the column order causes issues? This is really surprising to me.

[AnnaNzrv]

Fixed it!
The issue was caused by predict.rFerns, which attaches an extra attribute (timeTaken) to the returned factor. As a result, running prediction twice yields responses that are identical in values, class, and levels but differ in this attribute. That difference makes equality checks fail in the autotest (line 412):

all.equal(prediction$response, tmp$response)
#> FALSE  # because attributes differ