String interpolation revamp: design decisions

I am; the current implementation uses a new AST node instead of a closure. The pitch thread I was talking about, with a draft proposal detailing the design, has since been posted: