Thanks to Marino, I was shown that I was...wrong. And with math even, so it's not 'internet wrong'. It's just plain wrong.
My wife is laughing her ass off reading that sentence, by the way.
Using his ideas for a rotation and the published numbers I had for threat, he figured out (and I confirmed) that it is always better at any reasonable amount of AP/crit to keep up a lacerate stack for both maximum threat AND maximum damage. The threat gain is on the order of 5%, which is pretty good. It doesn't take into account the lacerate tax of having to do a bunch of lacerates early on in order to get the stack going, but that's not going to be huge in a longer fight.
The ideal rotation for a 36-second cycle (easier math) would be:
8 mangles (assuming imp mangle), 3 lacerates, 13 swipes, 14.4 mauls. You can substitute faerie fire for swipes in there as well if you want truly maximal threat. At 6250 AP, 40% crit and normal boss armor, we get 4344 TPS.
(if you're wondering why 6250 AP, it's because that's the number with normal raid boss armor reduction where swipe spam matches lacerate spam for threat).
For a cycle without lacerates and no lacerate stack, (so 16 swipes), we get 4090 TPS.
This holds true for AP up to 100,000 and above, in case you're wondering. It also holds true for 100% crit, as crazy as it sounds. The lacerate DOT is just that good.
What about a more reasonable, safe rotation that has a lacerate every 6 seconds instead of 12? That's still more - 4285 TPS. Not a huge loss.
Now, where it gets tricky is when you start considering how many lacerates it takes to keep up a stack. As you get better AP and crit, swipe starts overtaking certain numbers of lacerates. If you have to lacerate more than once every 6 seconds to keep the stack up, chances are that at higher levels of AP it won't be worth it. At really high levels of AP (we're talking post-WotLK levels) it takes one lacerate every 9 seconds to make it worth it.
So for optimal rotations, having lacerate in there is best. Fewer is better, but none is not ideal.
I think the next big mathy thing I'm going to do is figure out what the 'lacerate tax' costs you in threat vs swipe spam. The obvious view is that for shorter fights, getting a lacerate stack isn't worth it - but what is the cost of the lacerate stack falling off in longer fights? If you know for whatever reason that a lacerate stack will likely fall off (Malygos is an obvious one, but there are plenty of others), is it worth it to reapply the stack quickly? Slowly? This isn't the easiest answer to model.
Anyway, sorry that I was wrong. Thank you, Marino, for checking the math and asking the question. I always appreciate people doing that and callling me on it when I'm wrong, and I'm not so egotistical that I can't admit it.