predict type should be also visible in flow.name

[giuseppec]

OpenML computes the AUC also when the predictions are not probabilities and from
the list below, we don't really see that mlr.classif.randomForest(17) uses predict.type = "response" and mlr.classif.randomForest(21) uses predict.type = "prob".

res = listOMLRunEvaluations(task.id = 37)
res = res[grep("mlr.classif.randomForest", res$flow.name), ]
res[,c(1:5,8)]
#      run.id task.id setup.id flow.id                    flow.name area.under.roc.curve
#1297 1816243      37     9405    4782 mlr.classif.randomForest(17)             0.736701
#1612 1839540      37    24112    4849 mlr.classif.randomForest(21)             0.825437
#1613 1839541      37    24113    4849 mlr.classif.randomForest(21)             0.826119
#1614 1839542      37    24114    4849 mlr.classif.randomForest(21)             0.812451

One can only see this after

convertOMLFlowToMlr(getOMLFlow(4849))
#Learner classif.randomForest from package randomForest
#Type: classif
#Name: Random Forest; Short name: rf
#Class: classif.randomForest
#Properties: twoclass,multiclass,numerics,factors,ordered,prob,class.weights
#Predict-Type: prob
#Hyperparameters: 

convertOMLFlowToMlr(getOMLFlow(4782))
#Learner classif.randomForest from package randomForest
#Type: classif
#Name: Random Forest; Short name: rf
#Class: classif.randomForest
#Properties: twoclass,multiclass,numerics,factors,ordered,prob,class.weights,featimp
#Predict-Type: response
#Hyperparameters: 

My suggestions would be to modify the flow name so that one can see this info directly from the flow.name, e.g. something like mlr.prob.classif.randomForest(21).