Semantics of negative bounds

Constraints are supposed to be automatically normalized. But this does not appear to be the case when negative bounds are involved.

See e.g. the attached proof, which VeriPB 2 accepted.

It fails on an equality check of a constraint of the form

... >= 0

The internal constraint is the same, but of the form

... >= -xxx

Is this a new design decision or an oversight (since htese are constraints one is not really interested in anyway)