Returns the error dumps generated during resampling, which can be used with
to debug errors. These dumps are saved if configureMlr configuration
or the corresponding learner
The returned object is a list with as many entries as the resampling being used has folds. Each of these entries can have a subset of the following slots, depending on which step in the resampling iteration failed: “train” (error during training step), “predict.train” (prediction on training subset), “predict.test” (prediction on test subset).