Feature Selection Operators require same functionality as Optimization Parameters (ports, "logging")

All Feature Selection operators (specifically machinelearning:optimize_attribute_set_forward and machinelearning:optimize_attribute_set_evolutionary rn) require more ports (just like optimize parameters) and a logging-like output similar to the "iteration" port of Optimize Parameters.

• Flexible number of input ports to use information from outside within the FS
• Flexible number of output ports to retain information beyond the standard output ports
• Flexible number of loop-ports for non-parallel Feature Selections
• Output to show how the Feature Selection went, similar to "iteration" for Optimize Parameters. Specifically, we need one row per inner iteration, with these columns:
  • Numeric value(s) of the calculated Performance(s)
  • Numeric # of currently selected attributes
  • Nominal concatenation of selected attribute names
  • Some value to validate the stopping behaviour, e.g. the absolute / relative increase and / or the significance, depending on what is chosen