[REVIEW] Fix __repr__ function for preprocessing models

[viclafargue]

Closes #2963

[GPUtester]

Please update the changelog in order to start CI tests.

View the gpuCI docs here.

[wphicks]

Looks good! Could you add a test demonstrating the new functionality?

[wphicks]

LGTM!

[codecov-io]

Codecov Report

Merging #3191 (fbc9356) into branch-0.17 (f1abcb8) will decrease coverage by 0.46%.
The diff coverage is 28.00%.

@@               Coverage Diff               @@
##           branch-0.17    #3191      +/-   ##
===============================================
- Coverage        70.99%   70.53%   -0.47%     
===============================================
  Files              198      199       +1     
  Lines            15729    16207     +478     
===============================================
+ Hits             11167    11431     +264     
- Misses            4562     4776     +214     

Impacted Files	Coverage Δ
python/cuml/_thirdparty/sklearn/utils/_pprint.py	27.71% <27.71%> (ø)
...cuml/_thirdparty/sklearn/utils/skl_dependencies.py	52.89% <100.00%> (+17.35%)	⬆️
...ython/cuml/dask/neighbors/kneighbors_classifier.py	20.48% <0.00%> (-1.52%)	⬇️
python/cuml/dask/neighbors/kneighbors_regressor.py	30.00% <0.00%> (-1.25%)	⬇️
python/cuml/common/input_utils.py	94.14% <0.00%> (+1.60%)	⬆️
python/cuml/thirdparty_adapters/adapters.py	90.17% <0.00%> (+1.61%)	⬆️
...cuml/experimental/decomposition/incremental_pca.py	88.94% <0.00%> (+2.28%)	⬆️
python/cuml/decomposition/pca.pyx	94.60% <0.00%> (+2.40%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update f1abcb8...fbc9356. Read the comment docs.

[viclafargue]

I thought that it might be better to use cuML estimator Base class. It fixes the __repr__ problem. Unfortunately preprocessing is not fully compliant with this class yet. I'll create an issue to work on this. After second thought, I reverted the last commit as the usage of the cuML estimator Base class will require more time. I think that it can be done for the next release along with the exit outside of the experimental namespace.