I like using the array_constrain() because it separates the arguments. You have your formula that you are executing in the arrayformula() that is providing an output, and then you can define the constraints as a separate argument without complicating your main formula. Depending on how you want to constrain (every other row, last row regardless or gaps, etc) doing it all in one formula gets more complicated. I also think it's easier for people learning the arrayformula() techniques to see what the entire formula is doing by using the intended functions and visually separating what the two pieces do. The idea of working on your output, and then expanding by putting more functions around it. And, depending on the size of the data certain functions take longer. Query, lookups, logic statements, etc will take more calculation time then a count or a sum.
What the benefit vs doing arrayformula(if(a2:a="","",......)) ?
I like using the array_constrain() because it separates the arguments. You have your formula that you are executing in the arrayformula() that is providing an output, and then you can define the constraints as a separate argument without complicating your main formula. Depending on how you want to constrain (every other row, last row regardless or gaps, etc) doing it all in one formula gets more complicated.
I also think it's easier for people learning the arrayformula() techniques to see what the entire formula is doing by using the intended functions and visually separating what the two pieces do. The idea of working on your output, and then expanding by putting more functions around it.
And, depending on the size of the data certain functions take longer. Query, lookups, logic statements, etc will take more calculation time then a count or a sum.