style: rename user_token_number as code

This should have been done in 3.6, but I wanted to avoid introducing
conflicts into Vincent's work on counterexamples.  It turns out it's
completely orthogonal.

* data/README.md, data/skeletons/bison.m4, data/skeletons/c++.m4,
* data/skeletons/c.m4, data/skeletons/glr.c, data/skeletons/java.m4,
* data/skeletons/lalr1.d, data/skeletons/lalr1.java,
* data/skeletons/variant.hh, data/skeletons/yacc.c, src/conflicts.c,
* src/derives.c, src/gram.c, src/gram.h, src/output.c,
* src/parse-gram.c, src/parse-gram.y, src/print-xml.c, src/print.c,
* src/reader.c, src/symtab.c, src/symtab.h, tests/input.at,
* tests/types.at:
s/user_token_number/code/g.
Plus minor changes.

src/gram.h

@@ -173,7 +173,7 @@ typedef struct
 {
   /* The number of the rule in the source.  It is usually the index in
      RULES too, except if there are useless rules.  */
-  rule_number user_number;
+  rule_number code;
 
   /* The index in RULES.  Usually the rule number in the source,
      except if some rules are useless.  */
@@ -262,7 +262,7 @@ extern symbol **symbols;
    by the user's yylex routine, it yields the internal token number
    used by the parser and throughout bison.  */
 extern symbol_number *token_translations;
-extern int max_user_token_number;
+extern int max_code;