We showed a proof of concept of exactly this actually at our initial launch event for Framework Laptop 16 last year. It's still something we see as an interesting use case to support, but we don't have a timeline around productization of it.
Unsure how viable that would be, but it would help justify that proof of concept (imho).
Would be really cool if the GPU could serve multiple devices across generations like that somehow, eventually. Buy one FW laptop, upgrade it, use the old parts for a new handheld, use the old GPU as a dock for that handheld when you get a new one for the laptop.