Fix E-ACSL header date

See merge request frama-c/frama-c!3789

[e-acsl] fix error with recent GCCs in dev config

See merge request frama-c/frama-c!3783

restore coherence checks between API documentation and development guide

See merge request frama-c/frama-c!3479

[Variadic] avoid musl-related test differences

See merge request frama-c/frama-c!3769

[gui] Fixes a crash on the error dialog.

See merge request frama-c/frama-c!3767

[e-acsl] add quantification over enum types

See merge request frama-c/frama-c!3757

[e-acsl] Bugfix: register data in the presence of RTE

See merge request frama-c/frama-c!3668

avoid test differences in some ARM64 configurations

See merge request frama-c/frama-c!3768

Blindly accept all changes in doc and type of plugin-dev-guide-labelled
values since last time `make check-dev-guide` was successfully launched

the lexer and parser are broken beyond repair

references to constructors of sum type were not correctly taken into account

More specifically, preserve the underlying type name when analyzing type
constructors. Otherwise, correspondance with LaTeX index is lost.

Also do not forget to copy oracle in the `generated` directory

- clean up plugin files generated into lib/plugins after each run
- do not compile in dev mode (tutorial is not targeting a Frama-C dev env)
- ask ptests to abort in case of error, to make errors more visible

Actually tutorial in devman is not completely internal but uses bin/frama-c

Having relative directories in the PATH is rarely a good idea.
Moreover, if the PATH is set as part of the sub-make command, it does
not seem to always be given to the `$(shell ...)` command. It is thus more
robust to set it explicitely _before_ launching the sub-`$(MAKE)`

- making the dev guide (and especially its checks over the tutorial code)
should be done against up-to-date frama-c binaries.

- `Boot` module is referenced in the index of devman, it ought to be
present in the API documentation

`State_selection` has its signature embedded into signature `S`, shared
with submodule `Static`, marked as deprecated since 2012. If #1103 is
fixed, we can revert this commit.

With the current checks, indexing `M.s` where `s` is in fact
`include`d in `M`'s signature can't be done with
`\scodeidx{M}{s}`: you can only refer to `\codeidx{M}` and
`\scodeidx{I}{s}` where `I` is the module where `s` actually appears.

Changes seems to have taken place for 18.0-Argon according to the
comments on the relevant functions