stacking multiple successive preprocessing methods is not always helping

## Context of the issue

It is possible that TPOT proposes a model with 2 successive RobustScaler()
I have checked myself, this combination is not helpful.

## Possible fix

just a rule-based checking concerning these combinations
if F(F()) == F() then just keep F()
then if the new pipeline is actually already tested then it can be discarded or replaced by another one.

## `name of issue` screenshot


Start by tracing TPOT's pipeline-generation entry point for preprocessing operators and how RobustScaler is represented; the issue names no file or test. Done means successive equivalent preprocessing steps are collapsed and an already-tested resulting pipeline is not evaluated again, with coverage for this case.

Written by the indexing model from the issue text.