CV of conformal models

[manuelpastor]

So far, the CV of conformal methods is hardcoded to kfold. For classsifier it was further hardcoded to n=5. And the user WAS NOT INFORMED. This is a major problem

Now, at least the user was informed of the new settings and the value of n is read from the parameters

But... why hardcode to kfold????

[manuelpastor]

After testing with loo I understand the risk of using loo (veeery slow). However, this is a user choice and he must be informed of the settings.

[manuelpastor]

Fixed!