Consider passing the ID and position to the error-recovery function in the parser

[j-jzk]

(Instead of the whole ASTNode.)

It would be more consistent with the way the lexer does this.