(Again, sorry about the delay.)
This is a really interesting question. The concepts of “the higher physical productivity of roundabout processes” and that of “economies of scale” are definitely different things in terms of definitions, but you might be right that in practice, they have significant overlap.
They’re not identical, however. I think you might be right that any example of economies of scale would involve greater roundaboutness, but the reverse doesn’t hold. I can devote the same total number of labor hours to Direct Process A, which yields x units of output, as I do to Roundabout Process A, which yields (say) 2x units of output. So that wouldn’t be an example of economies of scale, because economies of scale says that when you increase the amount of inputs, the output rises more than proportionally. Yet in this case, the amount of input stayed the same, I just changed the degree of roundaboutness.
Now regarding the other main part of your question: There is the element of selection. People would already have chosen the most physically productive methods to produce their desired outputs, *except* in cases where they didn’t want to wait for a longer process to come to fruition. So in the original equilibrium, it will necessarily be the case that if the person is willing to wait longer, he can adopt a longer, more roundabout process that is more physically productive.
BB just assumes as an empirical fact that there always exist more roundabout processes that have a higher physical productivity (per unit input). But to repeat myself, at any point in time, humans have exploited the most productive processes they know of, that deliver the product in an acceptable timeframe. So on the margin, any *more* productive process *not* in use, is lying idle because it requires too long of a wait.
I spell this stuff out better in the first chapter of my dissertation.