This is one of those cases where the camera has a pretty clear answer to your question. So let’s look at the footage!
https://troygrady.com/seminars/antigravity/clips/descending-nines/
The descending nines clip from Antigravity is a pretty textbook example of down-up-rotate in practice. Watch this in slow motion. You’ll see that, for the most part, the rotational movements are paired with the string changes. Downstroke, upstroke, then a visible rotational movement and a switch to the next string. Then up, down, and a visible rotational movement to get back. We can debate how much the rotational movement is really spread out over the notes. But what is clear is that the most visible movements are tied to the string change note.
Here’s another example — fives across two strings:
https://troygrady.com/seminars/antigravity/clips/fives-2str/
The movements here are a little more subtle, but again, if you look at this in slow motion, you’ll see the most movement happens right around the moment of the string change.
But what do other players do? Let’s check out Teemu:
https://troygrady.com/interviews/teemu-mantysaari/teemu-mantysaari-clips/scale-sixes-descending-6str/
Pretty much the same thing. Watch this in slow motion, and you’ll see most of the movement happening around the string change. As in my case, there is some asymmetry - for a given pair of strings, the descending string change looks a little different than the ascending one. And this appearance changes depending on which strings he’s playing on. When he’s on the upper pair of strings, the ascending string change is the aggressive one, with more twist. When he’s on the middle pairs, the descending string change movement becomes more aggressive. This is likely due to anchoring. In other words, he’s not moving his hand placement to track the string in a one-to-one relationship, uniformly across the strings with the entire arm. Instead, it’s a mish mash of some arm placement change, some wrist placement change, some pickslanting changes. It’s complicated and mostly subconscious.
Ok fine let’s look at some upward pickslanting scale playing. Here’s Andy Wood:
https://troygrady.com/interviews/andy-wood/clips-guitar/bop-bach-and-roll/
There are many examples we can look at here, but taking a look at bar 12, watch the ascending string changes in slow motion. The pickslanting movements are again pretty much tied to the string change, but inversely. Andy is doing primary uwps here, so he reverts to dwps, and reverts back to uwps pretty quickly after the string change. You will see this quick compound switch and switch back movement everywhere in Andy’s playing. He’ll even do them back to back, which you could say becomes a crosspicking technique. Check out bar 5 of the same clip, the fourth through sixth notes of the bar.
Batio is another great example which we’ve looked at extensively in Antigravity, and we know that he does the same thing - a more aggressive movement tied closely to the string changes. And we can keep going with other players - lots of examples here.
The bottom line is that when you look at two-way pickslanters, the movements you see appear localized to the string change itself. Sometimes very quickly in a switch and switch back kind of way, as in Andy’s case. And sometimes split into halves as in my case. There may be some “spreading it out” happening in between these movements, we can debate. But the most aggressive movements happen at the string changes themselves.
Ergo, in your practice, if you feel like you’re doing more work around the time of the string change, that’s probably fine - that’s what it looks like many (most?) players are doing.
Edit: Didn’t really answer your question re: scenario 1 or 2. The answer is that both methods appear to work, and if you find yourself behaving more like scenario 2, that is fine. Andy and Mike are clearly scenario 2 players, and in fact, this is what we would thinks of as the classic case of having a “primary” pickslant, and using a compound movement to switch away from it, and switch back to it quickly. This approach appears to work fine at high speed.