fix future warning by avoiding importing private modules

[StrikerRUS]

Prepare for next scikit-learn version.

E:\Miniconda3\lib\site-packages\sklearn\utils\deprecation.py:144: FutureWarning: The sklearn.ensemble.forest module is  deprecated in version 0.22 and will be removed in version 0.24. The corresponding classes / functions should instead be imported from sklearn.ensemble. Anything that cannot be imported from sklearn.ensemble is now part of the private API.
  warnings.warn(message, FutureWarning)
E:\Miniconda3\lib\site-packages\sklearn\utils\deprecation.py:144: FutureWarning: The sklearn.linear_model.base module is  deprecated in version 0.22 and will be removed in version 0.24. The corresponding classes / functions should instead be imported from sklearn.linear_model. Anything that cannot be imported from sklearn.linear_model is now part of the private API.
  warnings.warn(message, FutureWarning)

[coveralls]

Coverage remained the same at 95.858% when pulling f085588 on future_warnings into d73048e on master.

[izeigerman]

Hey @StrikerRUS, thanks for this PR! Why didn't we limit the area of impact to those modules mentioned in the warning? Eg I don't consider sklearn.linear_model or sklearn.ensemble to be private. Please note that initially we were trying to follow the Google style guide on imports - http://google.github.io/styleguide/pyguide.html#22-imports

[StrikerRUS]

Now Powershell job is failing. Refer to #193 (comment) and #199 (comment).

=================================== FAILURES ===================================
_ test_e2e[xgboost_XGBClassifier - powershell - train_model_classification_binary2] _
estimator = XGBClassifier(base_score=0.6, booster='gblinear', colsample_bylevel=None,
              colsample_bynode=None, colsamp...ambda=0, scale_pos_weight=1, subsample=None,
              tree_method=None, validate_parameters=False, verbosity=None)
executor_cls = <class 'tests.e2e.executors.powershell.PowershellExecutor'>

...

expected=[0.04921597 0.950784  ], actual=[0.0492156801663393, 0.950784319833661]
expected=[0.06305683 0.9369432 ], actual=[0.0630565194153174, 0.936943480584683]
expected=[0.1270023 0.8729977], actual=[0.127001843824201, 0.872998156175799]
expected=[0.0785104 0.9214896], actual=[0.0785098971788289, 0.921490102821171]
expected=[0.8055698  0.19443017], actual=[0.805566852256706, 0.194433147743294]

[StrikerRUS]

Interestingly that expected output sometimes doesn't equal to 1.0 while actual one does.

I'm afraid that the root cause is #188 because I don't remember this error happened before that PR.

[StrikerRUS]

Thanks for the review!
I removed all additional changes in imports. Now this PR only directly addresses FutureWarning.