[eacsl] Add support for bitwise operators when using C integer types

See merge request frama-c/frama-c!2635

[eacsl] Add support to create GMP rational from GMP integer

Closes e-acsl#120

See merge request frama-c/frama-c!2700

[eacsl] Fix VLA memory tracking

Closes e-acsl#119

See merge request frama-c/frama-c!2715

[Eva] Eval_terms: fixes an error message.

See merge request frama-c/frama-c!2696

[Analysis-scripts] fix issues with relative paths in JCDB

See merge request frama-c/frama-c!2712

- Remove the renaming of Frama-C built-ins in the injector since the
  code printer is now responsible for it;
- Explicitely add calls to `store_block` (resp. `delete_block`) when
  allocating (resp. deallocating) a VLA.

Remove unused labelled argument and simplify the function as a result.

[eacsl:codegen] Support of variadic functions in `Constructor.mk_lib_call`

See merge request frama-c/frama-c!2706

fix handling of `loop assigns` by `Annotations.add_code_annot`

Closes #883

See merge request frama-c/frama-c!2702

it seems like part of the issue lies in the fact that the cache might be looked
at from different cwd. Anyways, proposing an absolute location can't hurt.

When loading some state, it may happen that some annotations are found to have
been emitted by emitters that are no longer usable. This commit let the kernel
adopt them, keeping them available as was the case before refactoring
`Annotations.code_annot` to use `Annotations.code_annot_emitter`.

code duplication started to become a bit problematic

All loop assigns corresponding to a given set of behaviors are now merged
together instead of being split into emitters. This is the same as what is done
for statement contracts and will make it much easier for callers to identify
appropriate loop assigns

[ivette] Property table: adds better tooltips to property and alarm kinds

See merge request frama-c/frama-c!2717

[Kernel] projectify option -add-symbolic-path

See merge request frama-c/frama-c!2719