Partial functions in sets: bug and semantics ambiguity

[Broes De Cat]

The behavior of partial function terms occurring within set terms does not coincide with the expected meaning: usually drop the term which have no mapping from the set. E.g. sum(...) > n and there is some term which has no mapping, then the closest atom is false. Additionally, the system crashes when partial function term occurring in an optimization term, as that code cannot handle terms without an image e.g. n/x with x=0 (which crashes in the associated file).

[Stef De Pooter]

When deriving bounds for a product term, do not add zero.

Fixes issue #249

→ <<cset 6f4bf3426f35>>

[Broes De Cat]

Not fixed by the mentioned commit.

[Broes De Cat]

Note: marc is thinking about how to integrate it in the rest of the language, we will probably add a "image-exists" predicate everywhere (but how to do this syntactically without repeating the term?).