glr2.cc: example: use objects (not pointers) to represent the AST

Currently we are using pointers.  The whole point of
glr2.cc (vs. glr.cc) is precisely to allow genuine C++ objects to be
semantic values.  Let's make that work.

* data/skeletons/glr2.cc (glr_state::glr_state): Be sure to initialize
yysval.
(glr_state): Add copy-ctor, assignment and dtor.
(glr_state::copyFrom): Be sure to initialize the destination if it was
not.
(glr_state::~glr_state): Destroy the semantic value.
* examples/c++/glr/ast.hh: Rewrite so that we use genuine objects,
rather than a traditional OOP hierarchy that requires to deal with
pointers.
With help from Bruno Belanyi <bruno.belanyi@epita.fr>.
* examples/c++/glr/c++-types.yy: Remove memory management.
Use true objects.
(main): Don't reach yydebug directly.

* examples/c++/glr/local.mk: We need C++11.

TODO

@@ -9,6 +9,23 @@ Clarify that rule numbers in the skeletons are 1-based.
 There are many macros that should obey api.prefix: YY_CPLUSPLUS, YY_MOVE,
 etc.
 
+** YYDEBUG etc. in C++
+Discourage the use of YYDEBUG in C++ (see thread with Jot).  Stop supporting
+#define YYSTYPE by the user.
+
+Add value_type as a synonym for semantic_type.
+
+** Asymmetries
+In glr_state, we have yysval and yylloc.  It should be yyval/yyloc (and
+yylval/yylloc when referring to the lookahead).  glr.c should
+s/yysval/yyval/.
+
+Also
+
+    yystack.yyglrShift (create_state_set_index(0), 0, 0, yylval, &yylloc);
+
+Why are yylval and yylloc treated differently?
+
 ** yyerrok in Java
 And add tests in calc.at, to prepare work for D.