>     I don't think that's a great API, it makes it impossible to download
>     files that are larger than the amount of RAM currently available.
>
> That's part of why it's called "simple".  But I suppose there does need
> to be a way to provide piece-by-piece handling in the main API.

 From experience, offering the response body as a byte input port highly
recommended.

R6RS has `get-bytevector-all` (among others). Hence in throwaway
scripts, you could just do something like:

(utf8->string (get-bytevector-all (response-body response)))