Okay, I’ve been telling people that’s not feasible, but let me explain why.
Yes, you can grab the backup of the game database before the thrall decay snafu. Let’s call that one the pre-snafu database. And you can also look at the event log in the post-snafu game database.
With that, you might be able to come up with a query (or a script) that will give you the diff. Let’s suppose that you can. I’m not sure, but let’s assume for the sake of discussion that there actually exists a way to solve this non-trivial problem.
Now that would give you a list of thralls (and their inventories and stats and gear) to “resurrect”. Where do you put them? In their original locations? What if the player built something there while you were off working on this solution? What if the player’s stuff is gone from there (e.g. wiped by the alpha) and some other player built their stuff there? What if the player has been adding followers to compensate for this fiasco and now resurrecting the lost ones puts the clan over the hard limit and the system starts killing off the excess followers at random?
When I say “this shіt is hard”, I’m really not trying to exaggerate 
“It wasn’t sufficient” is a reasonable-sounding guess, but it’s a guess. We’ll never know. The only way we could know would be to grab the database from one of the officials, put it on a test server of equivalent capacity, get an equivalent number of players to play “like normal” on it, and then experiment with different thrall populations. Something along these lines:
- Killed all but 100 in each clan. Still performs like shіt.
- Killed all but 50 in each clan. Slight improvement.
- Killed all but 30 in each clan. Noticeable improvement.
- Conclusion: 30 should be the limit.
As for it being mixed with all sorts of other factors, that’s exactly my point. I’ve already made the mistake of assuming I know enough about performance based only on my observations of the black box this game is. I was proven wrong and I’ve learned my lesson. Now I’m trying to get others to do the same 
Your experience and my experience and every other player’s experience is a starting point. The game remains a black box and reasoning about it is tricky.
Okay, that’s a start. What happens then? Does your performance keep sucking until they log off? Or until they move out of the render range? Or does it get better a couple of seconds later? Don’t take this as an attack or criticism. Those are just some questions that someone who’s actively trying to diagnose this stuff would have to ask. I’m mentioning them to explain why “having 200 thralls is a performance problem” is not a thing anyone here can use to justify suggesting measures to deal with that.
I’m not saying that merely having 200 thralls isn’t causing any performance problems. I’m saying that it might or might not be, and if it is, we don’t have enough knowledge or insight to suggest how to deal with that.
Oh, and that mention of the lights? That’s precisely why I keep saying this stuff is hard. There’s no good reason I can think of why the lights would have any special effects on the server. On the other hand, sending the information about a big number of placeables to the client could make the server lag while it’s doing it, but that would be true for any kind of placeable, not just the lights. I guess using torches is the easiest way of accomplishing that, so that’s why people talk about the lights so much. And it has the added “benefit” of lagging your client, too.
Sorry, this has gone way off the original topic, but it’s actually more interesting (to me) than the whole squabble about whether rollbacks were okay and whose play time matters more.
No doubt people would flip their shіt if someone said “Hey, we should start reporting people for having too many thralls and hopefully Funcom bans them.”
I do agree with you that if you can get people banned for clustering 200 torches in a certain pattern, regardless of whether they did it maliciously or just because they don’t give a shіt about anyone else, then you should be able to get them banned for clustering 200 AIs in a pattern that produces similar effects.
That doesn’t mean I agree with how Funcom is handling any of that in their enforcement of the rules.
While I do have respect for your experience in PVP, I also have to be skeptical, because people have claimed some weird things about thralls over time.
Case in point: there was a thread asking to nerf horses’ health, of all things, partly because of the issues with the AI. The reason I’m bringing that up is that, if it’s true, it highlights just how far PVP players will go to gain an advantage.