Update Version 1.0.1

[suecharo]

Update Version 1.0.1

[suecharo]

1. Added the Gram matrix normalization
   Thanks to this modification, the HSIC score takes [0 1].
   More specifically, let us denote K as the normalized Gram matrix by its Frobenius norm.
   Then, trace(K*K) = 1.

   Due to the modification, I also slightly modified the selected features list
   in test_classification.py and test_+regression.py to pass the tests.

2. Modified the README.md
   Included the following lines:

   >>> from pyHSICLasso import HSICLasso
   >>> hsic_lasso = HSICLasso()

This would help users to use our package.

3. Sorting A with respect to feature importance (beta)

4. The feature importance of the last selected feature was zero. Thus, I fixed.

5. Due to sorting A, I rewrite the unit tests.