Harness domain properties

[triska]

Currently, dif/2 is sometimes used in cases where (#\=)/2 can propagate more strongly.

This also affects if_/3, where plain unification is currently used instead of (#=)/2. The predicate zcompare/3 allows to use if_/3 with CLP(FD) constraints to the full advantage of both.

[JCumin]

@triska I have replaced dif/2 in divisible_by/2 with #\=. I don't see any other part of the code where there is this issue.

As for if_/3, I am not sure I understand what you mean. Do you mean that there should be a (#=)/3 in the if part instead of (=)/3, or that I should use (#=)/2 in the then part instead of (=)/2?

If it's the former, I imagine you have already implemented a reified version of (#=)/3 yourself?
If it's the latter, I didn't think there was any difference between using (=)/2 and (#=)/2 when one of the sides is a ground expression.

[triska]

Here is one example of using zcompare/3 to implement (#<)/3:

#<(X, Y, T) :-
        zcompare(C, X, Y),
        less_true(C, T).
less_true(<, true).
less_true(>, false).
less_true(=, false).

The other arithmetic relations can be easily defined analogously. This makes clear that the arguments are integers, which is sometimes useful (for example, to obtain suitable type errors), and needed in more general cases.