FWIW, from this related thread, vcat allocates but seems to be faster than Iterators.flatten.
vcat
Iterators.flatten