On the contrary, I think that the left piece of code is not building constrains prematurely and actually enables you to modularize it away when needed.

Sure, if the logic grows, if it needs to scale, if the team increases in size... then it makes sense to modularize it. But building something from the very beginning to achieve that is going to impose constraints that make it harder to reason about and harder to refactor; you'll have to break down the previous structures and boundaries built by the function heavy example, which will probably introduce needless indirections.