Should we have comments in BNF?

PsiBuilder ignores whitespace and comment tokens but we have to use comment token in BNF just to have corresponding IElementType generated.

Even the official documentation for [Custom Language](https://plugins.jetbrains.com/docs/intellij/grammar-and-parser.html#define-the-grammar) has it as example: `private item_ ::= (property|COMMENT|CRLF)`.

In my own grammar I've ended up with an 'orphan' rule and it works fine: `private comment ::= COMMENT`

If comment tokens doesn't add anything to a parser, shouldn't we stop using them as we do for whitespace tokens. Of course they are necessary for corresponding IElementType generation but we could use an 'orphan' rules for them or even have a predefined `com.intellij.psi.TokenType`.

Start by reviewing the Grammar-Kit handling of BNF comment tokens and the IntelliJ grammar-and-parser documentation linked in the issue. Compare the documented `COMMENT` usage with the proposed orphan rule and predefined `TokenType` alternatives; done means an agreed direction for comment-token handling is documented or implemented.

Written by the indexing model from the issue text.