Cleans up and saves partial results on sigint signal.
Signal handlers are now registered at the start of an analysis, and previous
signal handlers are restored at the end.
New function [protect f ~cleanup] in value_util, that runs [f] and applies
[cleanup] in case of a user interruption or a Frama-C error, only if option
-save is set.

- [cacheable] is given at the registration of a builtin, and not separately
  for each result of the builtin.
- Removes the offsetmap for the arguments: no builtin were using it anyway.
- Returns a Cvalue.V.t for the result instead of an offsetmap.
- Builtins can also return:
  + only the value of the result, in which case the post-state from the
    interpretation of the specification is used with the resulting value
    computed by the builtin.
  + only the post-states for functions with no froms dependency (other than the
    result and arguments) and no write of local variable addresses.

Simplifies some builtins accordingly.

This annotation sets the maximum number of subdivisions in the evaluation
of expressions at a given statement.

Overrides the global option -eva-subdivide-non-linear for the given functions.

- shares and moves the functions [reduce_offset_by_validity] of Locations and
  Precise_locs into base.ml.
- replaces the boolean argument ~for_writing into the new type access, that
  represents Read, Write or No_access. Without any access, offsets must point
  into or just beyond the base validity.
- fixes the support of accesses of size 0: they are now invalid:
  + in bases with Invalid validity;
  + one past a base validity unless the base ends with an empty struct.

(blind make headers from specifications)

- Fix Value/Value#5
- Fix Value/Value#14

Some case studies (e.g. dyad) use some ugly casts from fd_set_t which lead
to the analysis stopping too early.
Changing the representation of fd_set_t should also help it better conform to
the standard (since a fd_set_t should be able to hold FD_SETSIZE elements).

[Ptests] preserve LOG after STDOPT directive

See merge request frama-c/frama-c!2073

This is absolutely not a sneaky attempt to relaunch a build (now that OCI seems
in better shape) pushing a nearly empty commit.