[Feature] Lexical modes

[oxc]

It would be nice if the library supported lexical modes (at least that's what ANTLR4 calls them: https://github.com/antlr/antlr4/blob/master/doc/lexer-rules.md#lexical-modes), where the lexer keeps a stack of modes and produces different tokens based on the current mode.

For example, I have a small mini language that basically specifies boolean expressions. Now I would like to use these expressions inside a larger document language, where they can occur only in certain positions in the header of the document (after a "@require" token until the end of the line). It would easily be possible to reference the root parser of the mini-language inside the document-grammar, but since the tokens are not compatible, this fails. I can workaround this by specifying a token that matches the whole line, and parse this separately in a second step. This, of course, is very hacky, and does not work in all situations.

It would be much nicer if I could switch the lexer to the mini-language mode when encountering a @require token, and then popping back to the default mode when encountering the EOL token. Ideally, I could "import" the mode from another Grammar. ~~Perhaps it might even make sense to make this the preferred way to specify different modes.~~

Let me know if I should create a minimal example.