bug: typedef is not treated as a general specifier

Did you check existing issues?

• I have read all the tree-sitter docs if it relates to using the parser
• I have searched the existing issues of tree-sitter-cpp

Tree-Sitter CLI Version, if relevant (output of tree-sitter --version)

tree-sitter 0.25.3

Describe the bug

In this grammar, typedefs are considered specific standalone statements, requiring the statement to start with typedef, but in the official C++ grammar, typedef is just one of many possible decl-specifiers.

This means that while e.g. typedef int x, y; is parsed correctly, int typedef x, y; is not, despite both compiling to the same effect in gcc and clang.

This is actually used in the wild, but it seems very rare. Nonetheless, I thought I'd report it in case someone's willing to bother themselves with strange and unknown C++ quirks :)

Steps To Reproduce/Bad Parse Tree

Output for int typedef x, y;:

(translation_unit [0, 0] - [1, 0]
  (declaration [0, 0] - [0, 17]
    type: (primitive_type [0, 0] - [0, 3])
    declarator: (identifier [0, 4] - [0, 11])
    (ERROR [0, 12] - [0, 13]
      (identifier [0, 12] - [0, 13]))
    declarator: (identifier [0, 15] - [0, 16])))

Expected Behavior/Parse Tree

The expected parse tree is the same as with typedef at the start, only with different locations:

(translation_unit
  (type_definition
    type: (primitive_type)
    declarator: (type_identifier)
    declarator: (type_identifier)))

Reproduce the issue by parsing typedef int x, y; and int typedef x, y; with the tree-sitter-cpp grammar, then inspect the declaration and declaration-specifier rules that distinguish these forms. The work is done when both inputs produce equivalent type-definition parse trees without an ERROR node.