* Nested comments are supported
          * Add a special exception for the (*) token

[trywhy3] Allow one to compile trywhy3 with debugging information and source map file by passing DEBUGJS=yes to make.

- Define set with create (and constrained a benign mistake on axioms)
- s/FloatMatrix/Matrix

to handle non-square matrices

This patch solves a problem on the order of the generation of new
declarations by intro_projections_counterexmp. The problem came from
the combination of Trans.on_tagged_ls and Trans.decl. The former was used
to know each projection function in advance and the latter would
introduce new declarations directly at the definition of the decl being
analysed sometimes using projections functions before they are defined.

This patch makes intro_projections_counterexmp.ml adds declarations only
at the end of the task. So we are sure that any meta is really defined
before.

* src/transform/intro_projections_counterexmp.ml
 (introduce_constant): Minor clarification.
 (projections_for_term): Replacing append with proper symbol. Also adding
  already computed list of declaration to the output.
 (introduce_projs): Removing already defined declarations from the set of
  declaration we will add.
 (build_projections_map): The order of elements is irrelevant so
  it is faster to use cons instead of append.
 (meta_transform2): Add declarations created by f at the end of the task.
 (encapsulate): Compose meta_transform2 and introduce_projs.
 (commented meta_transform2): Request for a fold able to do what is
  described.

The one liner is missing because we did not project value of array
elements in intro_projections_counterexmp. I changed the whole algorithm
of projections to allow projecting values of multidim array. Also,
moved local functions outside the main function. Added some trivial
helping functions.

* intro_projections_counterexmp
 (detect_map_types): Takes a type and return the list of successive types
  in the map and the return type. On map int (map int int) returns
  ([int; int], int).
 (last_type): Takes a list of proj_functions and returns the return type
  of the last projections_function that will be applied.
 (recreate_types): Recreates the type of the map with projected return
  type.
 (create_index_list): Returns a list of new well typed symbol for
  application in the axiom.
 (recreate_term_applications): Apply array to indices.
 (list_projection_until_base_type): Returns the list of all possible
  list of proj_functions applied to it.
 (projections_for_term): Changed the map part almost completely. We do
  not call this function recursively but we used other functions to get
  the recursive behavior.

To install the Why3-related Vim files, just create a symbolic link:
  ln -s "$(why3 --print-datadir)/vim" ~/.vim/bundle/why3

Thanks to Johanness Kanig for the suggestion.

Change-Id: I296e85a9aa76594b51bf045f61df65f2e20e3a35

Sometimes the windows syscall GetExitCodeProcess returns a large result,
larger than the ocaml [int] type. This is expected and not an error. For
example the constant STATUS_QUOTA_EXCEEDED, which is 0xC0000044, may be
a valid exit code. Such large values trip up the [int_of_string] parsing
in the client.

This patch implements the following solution to this problem:
* the server doesn't care and sends the large value;
* (the server now recognizes this value, though, to set the timeout flag
  more often)
* the client uses an Int64 value to parse that big constant;
* when converting to the internal Unix.process_status type, we simply
  convert to [int], because such large values don't have any special
  meaning for Why3 anyway.

* call_provers.ml
(parse_prover_run): now directly take the exit status as argument, and
convert it to int;
(handle_answer): don't wrap argument to parse_prover_run into unix type
* prove_client.ml
(read_answer): read Int64 type now
* server-win.c
(handle_child_event): set timeout boolean also when exitcode is equal to
   constant STATUS_QUOTA_EXCEEDED

Change-Id: I1163a6f1adf1bdbfe1f53269ce0ae57dc8bd0287

Contribution of Nicolas Jeannerod [niols@niols.fr]

- Call_provers.parse_prover_run does not attempt fixing answer anymore,
  except in the case where the answer is HighFailure and time is close
  to time limit (which is considered as Timeout)

- Session_scheduler.fuzzy_proof_time is now more liberal, accepts
  that two answers Unknown or Timeout of OutOfMemory with less than
  10% difference in time are equivalent, and thus should not be
  reported as a significant change

This commit also underspecifies the sign of product and quotient in case
of NaN result, so as to be consistent with the single NaN of smtlib.

We changed t_map_simp, track_values and the eval_match transformation
in order to prevent them from removing terms whose head has label
 keep_on_simp. Note that simplification inside those terms is
still possible.

* src/core/term.ml
(t_map_simp): Adding the case when f has label keep_on_simp.

* src/transform/eval_match.ml
(eval_match): Adding keep_on_simp as a stop label.

* src/whyml/mlw_wp.ml
(track_values): Stopping on keep_on_simp label.