How to randomize two blocks but spaced apart for within subjects?

I think what you want to do in this case is, in the survey flow), to use a combination of Randomization, Embedded data, branch logic, and duplication of blocks.
First, you'd want to put a randomizer step at the beginning to assign each respondent to an "order" (so you have 2 embedded data options, for example, creating an embedded data field called "Order" and using the randomizer step (show 1) to set Order to "AdaptiveFirst" or set Order to "ClassicFirst")
OK, so now your respondents have a randomly assigned order, and you have a means at the other end to know which order everything was presented.
Now you have a branch set. where you want the split experience to begin. If Order=AdaptiveFirst, show the Adaptive block; if Order=ClassicFirst, show the Classic block.
Then show all the other blocks.
Then later, where the "other" (previously unseen) block should be, duplicate the Adaptive and Classic blocks from above, and use branching again, but this time reversing the logic: If AdaptiveFirst, show the Classic block, if ClassicFirst, show the Adaptive block.
I hope that helps!