Dependency fix

[saattrupdan]

Updated the imports to their current versions. Specifically:

1. sklearn.externals.joblib is replaced with joblib
2. sklearn.utils.testing is replaced with numpy.testing
3. sklearn.externals.six is replaced with six
4. sklearn.tree.tree is replaced with sklearn.tree
5. presort parameter removed from calls to DecisionTreeRegressor
6. Replaced occurences of assert_true, assert_false, assert_greater and assert_less with their equivalent assert statements
7. As ForestClassifier and ForestRegressor can't be imported from sklearn anymore, the source code for these are now placed in forest.py

All unit tests in tests/, mondrian/tree/tests/, mondrian/ensemble/tests and quantile/tests run with no errors.

[Bougeant]

Is there a chance we can land this? I'm happy to review if needed.

[Bougeant]

Looks good to me

[PSSF23]

Thanks for making it work again! Really appreciate it!

[PSSF23]

@MechCoder It will be great if this PR could be merged!

[lnpsiqueira]

Hello,
Did anybody got this error?

rfqr = RandomForestQuantileRegressor(random_state=0)
Traceback (most recent call last):

File "", line 1, in
rfqr = RandomForestQuantileRegressor(random_state=0)

File "C:\Users\HerrSiqueira\anaconda3\envs\mestrado\lib\site-packages\skgarden\quantile\ensemble.py", line 295, in init
base_estimator=DecisionTreeQuantileRegressor(),

File "C:\Users\HerrSiqueira\anaconda3\envs\mestrado\lib\site-packages\skgarden\quantile\tree.py", line 222, in init
super(DecisionTreeQuantileRegressor, self).init(

TypeError: super(type, obj): obj must be an instance or subtype of type