the wesl grammar currently recognizes line and block comments with preParse(). It's nice to recognize comments with a grammar. And having a grammar was powerful for an earlier version of the language that included semantic information inside comments. But now that the language is simpler, it'd be faster to recognize comments in the tokenizer.

• Add a hook to the tokenMatcher to allow parsing entire comments into tokens. (Perhaps invent an api to pass a user provided function to the tokenMatcher..)
• Add those comment tokens into the matchingLexer ignore set
• profit!