Incorrect table potential after deleting an arc in the DAN of the dating problem

[Manuel Luque]

If we open in the GUI the DAN of the dating problem and we delete the link "NClub?" -> "Club", then the probability table of "Club" appears containing only zeros. This is not correct, because the sum of the probabilities should be equal to 1.0.

I attach the DAN of the dating problem.

[Jorge Pérez]

Temporal patch applied to check if the CPT sum is zero (in this case we replace the potential with a >uniform distribution).

It should be mandatory to implement the method "validate" for all potentials (for example, by forcing the method to be abstract in the class "Potential"). Also it must be revised how to update the potentials (if possible) and how to inform the user that the potential will be changed. I will open two new "issues", one for each task.