[Feature]: Support guided decoding with multistep decoding

[joerunde]

🚀 The feature, motivation and pitch

See #8985. It would be great if we could get the speedup from multi-step decoding without having to disallow users from using guided decoding.

I have no idea how feasible that is to do, but if anybody has a sketch of how it would be done I could be up for learning and helping to implement. I'm mostly opening this issue so it's documented and I can link it from the feature compatibility matrix in the docs.

Alternatives

No response

Additional context

No response

Before submitting a new issue...

• Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.