(that is, using number of occurrences)
No more definition of permutation using inductive predicates.
Impacts array.ArrayPermut; proof sessions updated.
Coq realizations for map.Occ and map.MapPermut;
proof session for array.ArrayPermut in progress

Problem: lambda is not extracted into a fun ... -> ...

    in order to be more platform independent

(and less axioms)

that proof was started during the preparation of the second VScomp,
but never completed

Will be supported in programs as soon, as we enable lambdas.

Since we are going to handle partial application of logical symbols,
there is no need to be rigid about currying: (((=) 0) 1) should be
(0 = 1) and not (((\ x y. x = y) @ 0) @ 1).

We do not want to generate distinct liftings for every single
partial application of a function or predicate symbol. Canonical
closures have an easily recognizable shape, thus we can detect
them and replace them with a unique constant "f_closure".