You can’t just throw a lot of questions into a hat, mix them up, and then count out enough to make a test. You have to have so many level 1’s, 2’s, etc. And then one passage on science, one in fiction, two parallel structure questions, four scatterplots of different kinds, etc. Getting the right combination would be some work.
Further, the order in which a question appears affects its difficulty. The same question later in a test is “harder” (fewer students answer it correctly). The difficulty is also affected by what questions came before.
So the remixed tests would have to be equated.
It would not be a simple procedure.