Having by chance read the older indirect for structs thread the @FixedArray/"Views over fixed-size storage" and @Erik_Eckstein points make me wonder if this could or would be abused to define recursive structs, like trees and such.
As that thread mentions people have used the regular non-fixed arrays to get a level of indirection to store values of a struct inside itself (heap be damned, I guess). I've done it myself, actually, but it feels icky as you're using a non-fixed array in spite of knowing how many elements it can/should always have (in a simple linked list either one or none).
Or in other words: If there is something like a fixed array, would this be a valid and good place to use one? Because people will surely do that.
Now I don't know much about what move-only types would entail, so I am not sure whether this potential backdoor into quasi-indirect structs would be even possible with the approaches discussed here (i.e. if there are aspects of this that would be preventing an indirection/recursion), but I thought I'd bring that up so smarter folks can think about it. 