[ENH] Train/Validation functions

[samihamdan]

In some situations it is needed to only perform a normal training or simple (not nested) cv on a dataset. In others more frequent ones it will be important to be able to validate a model on new test data.
These functions should have the similar interface like run_crossvalidation to make the use of julearn intuitive and consistent.

[fraimondo]

Can we make it to 0.3.0 with this feature?

[fraimondo]

Implementation:

for the fit function, we need to encapsulate the firsts lines of code of run_cross_validation up to the check_cv function. Then we just need to call the pipeline with the fit method instead of cross_validate.

On the other hand, the predict function (which I consider to be the most needed function), needs to have as input the same data as run_cross_validation:

• X
• data