Add code for utilizing GPU

[mainguyenanhvu]

I add one parameter for using TensorFlow Decision Forests (TF-DF). This package can help us to utilize GPU when we install tensorflow-gpu.

I believe that it could enhance the performance of ppscore when we have a large number of features by making use of GPUs.

[FlorianWetschoreck]

Thank you a lot for your PR! I will have a look at it in the new year.

[mainguyenanhvu]

I run test_calculation.py for my code. However, it did pass test for random_seed. I do not know why even when I printed the results to check. It illustrated they were the same.

python3 tests/test_calculation.py 
WARNING:root:TF Parameter Server distributed training not available (this is expected for the pre-build release).
{'x': 'x', 'y': 'y', 'ppscore': 0.6554396626978072, 'case': 'regression', 'is_valid_score': True, 'metric': 'mean absolute error', 'baseline_score': 1.0074503071153915, 'model_score': 0.347127417634877, 'model': DecisionTreeRegressor()}
{'x': 'x', 'y': 'y', 'ppscore': 0.6554396626978072, 'case': 'regression', 'is_valid_score': True, 'metric': 'mean absolute error', 'baseline_score': 1.0074503071153915, 'model_score': 0.347127417634877, 'model': DecisionTreeRegressor()}
False
Traceback (most recent call last):
  File "tests/test_calculation.py", line 335, in <module>
    main()
  File "tests/test_calculation.py", line 326, in main
    test_score()
  File "tests/test_calculation.py", line 176, in test_score
    df, "x", "y", random_seed=1
AssertionError