On N127-001, an optimisation of the fast WP was introduced, based on
marking states that contain variables that need to be refreshed when
merging them. However, this "marked" status was incorrectly reset to
false when merging states, even though some of those "unfresh" variables
could survive the merge. The consequence was that the WP was reusing
variables too aggressively, which would result in incorrect VCs.

We now compute the marked status of a merged state from the states to be
merged: if one of them is marked, the merged state is marked too.

mlw_wp.ml:
(merge): synthetize marked status from input states

Change-Id: Ifb02c54fca58137c31762469e48b59e7c907b995

- most important improvement : the explanation, used as first part
  of the shape was computed twice ! (at least)
- several improvements in the implementation, avoid using Format
  in particular (done also in checksum computations)
- possible further improvements:
    do not compute shapes if checksums are OK
       (they must be the same as in previous session)
    do not compute goal checksums if theory checksum is OK
       (they must be the same as in previous session)

If drivers in why3.conf are simple names like "alt_ergo", then the driver file
is search as <datadir>/drivers/alt_ergo.drv

This behavior is now the same as when a driver is given with option -D on the
command line for why3prove, why3replay or why3extract

Reminder: the datadir is either given as
1) the environment variable WHY3DATA
2) the field "datadir" of the [main] section of the
   why3 config file if exists
3) or by default the compile-time datadir

The underlying Monoid does not need to be commutative,
and is indeed not in example fibonacci.mlw

(so that we can use it with 'why3 execute')

For every check, we now print the "check_tree" into the JSON output.
This is basically a dump of the session tree, just in a different
format.

* json.ml
new json datatype and new function "print_json" to print it

Change-Id: I0b28ac181793af4d160abc58c2634b3cbcac9b94

Parsing reference was forgotten in commit for Q217-025. This solves the
problem with a new token MK_ANYTHING.

* src/driver/parse_smtv2_model_lexer.mll
(parse): Added lexing for all remaining mk*___ record constructs.

* src/driver/parse_smtv2_model_parser.mly
(smt_term): An smt_term can be a reference to a smt_term. It means it can
be (Mk_anything smt_term).

Change-Id: Ib529fee83d16d09266e526ca5fb3c65526addff5

Added a counter for constructors. We print a new functions to redefine
constructors only if there is a need.

* src/printer/smtv2_cvc_ce.ml
(print_constructor_decl): Added counter.
(print_data_decl): Added counter.
(print_saved_constructors): Print only if number of constructors in the
list is greater than counter.

Change-Id: Iad0f29caac961644dcbf4137341abd76c01e1090

Do not print (get-info :reason-unknown) on smt2 output. Previously,
detection of end of model and beginning of statistics was done by using
the reason. We now detect a single parenthesis on a line: every cvc4 model
ends with this (we assume statistics does not contain a line beginning
with a closing parenthesis). As a side effect, we also find
more counterexamples.

* src/driver/parse_smtv2_model.ml
(parse): Revert changes that were detecting Abort trap.

* src/driver/parse_smtv2_model_parser.mly
(output): Revert changes to allow absence of parenthesis at the end. Also
removed parsing on results given by the reason. Now match on ).

* src/printer/smtv2_cvc_ce.ml
(print_prop_decl): Remove (get-info :reason-unknown).

* src/printer/smtv2.ml
(print_prop_decl): Remove get-info for coherence.

Change-Id: Ifbd7b2b9ce499f8aaa020f58375d784793990965

This patch allows parsing model including the error message raised by cvc4
on mac.

* src/driver/parse_smtv2_model.ml
(parse): Allow parsing of specific error message "Abort trap".
(output): Allow parsing when last parenthesis is missing which is the case
when the bug is occuring.

Change-Id: I80d7276d37916e09ba4bb28478ac9a7427789771

Change-Id: Id165717756375db150031589740470cc1e08b8ac