Problems with constructed types

[Ingmar Dasseville]

The file below is not yet working after my propagatorfix. There still are some problems wit hconstructed types. We should check if pull request #175 fixes these issues...

[Bart Bogaerts]

See #616

This fixes a problem with twovaluedness of propagators. The problem was that during creation, twovaluedness of function symbols was used, while at the end of the process, twovaluedness of PFSymbols. They are not treated the same. I extended the initial check to also take into account the graphed versions, and in the mean time, generalised the code.

In fact, this failing assertion was not even a real bug; code was correct without the assert. The assert simply is an approx check that structure did not change. Now, it should always work

→ <<cset 5741be605aba>>

[Bart Bogaerts]

Resolved on Broes's branch (constr_...)

Now throws a proper error message about problems with infinity

[Broes De Cat]

Merged in propagatorfix (pull request #176)

See #616

→ <<cset 6182473b909a>>