mlr3pipelines 0.7.2

• New parameter no_collapse_above_absolute for PipeOpCollapseFactors / po("collapse_factors").
• Fix: PipeOpCollapseFactors now correctly collapses levels of ordered factors.
• Fix: LearnerClassifAvg and LearnerRegrAvg hyperparameters get the "required" tag.
• New parameter use_groups (default TRUE) for PipeOpSubsampling to respect grouping (changed default behaviour for grouped data)
• New parameter new_role_direct for PipeOpColRoles / po("colroles") to change column roles by role instead of by column.
• Dictionary sugar functions po() / pos() / ppl() / ppls() now make suggestions for entries in both mlr_pipeops as well as mlr_graphs when an object by the given name could not be found in the respective dictionary.
• New PipeOp PipeOpDecode / po("decode") to reverse one-hot or treatment encoding.
• Fix: Columns that are feature and something else no longer lose the other column role during training or predicting of PipeOps inheriting from PipeOpTaskPreproc.
• Fix: Made tests for PipeOpBLSmote deterministic.
• Fix: Corrected hash calculation for PipeOpFilter.
• New PipeOps PipeOpEncodePLQuantiles and PipeOpEncodePLTree that implement piecewise linear encoding with two different binning methods.
• Compatibility with new R6 release.
• Docs: Performed cleanup and standardization.
• Docs: Performed cleanup of reference index page on website.
• Docs: Fixed parsing of examples on website for PipeOpNMF and PipeOpLearnerPICVPlus.
• Fix: PipeOpTargetMutate and PipeOpTargetTrafoScaleRange no longer drop unseen factor levels of features or targets during train and predict.
• Simplified parameter checks and added internal type checking for PipeOpTargetMutate.