[translation] add a missing cast when an integer is used in a floating point/real context (was revealed by a Value's change, and now reveals a bug with == over reals)

[tests] update oracles according to various recent kernel changes

[E-ACSL] fix typing bug with divisions/modulos over floats

[E-ACSL] improve ACSL contracts of its libraries
[E-ACSL] no more unknow locations in translation of quantifications

[E-ACSL] my previous tentative patch did not work. I prefer to have E-ACSL that works than non-regression tests...

[E-ACSL] fixed bug with global declarations involved in the memory model (btw fixed but #1392)
[E-ACSL] additional feedback messages

[E-ACSL] e_acsl_assert now takes the crashing function in argument

[E-ACSL] code of e_acsl_assert put in e_acsl.c (no more in e_acsl.h)

[E-ACSL] convert postcondition even in presence of untranslated assigns clause
[E-ACSL] fix bug when the translation of a predicate containing \at is abruptely stopped because of a not-yet-implemented construct (the generated code contained undeclared __e_acsl_at variables)

[E-ACSL] prevent runtime errors in annotations (all of them)
[E-ACSL] fixed bug in interpretation of \valid(\at(...))
[E-ACSL] fixed bug in interpretation of locals in blocks involved in memory model