How on earth do you still get ranking for words and phrases that no longer exist on your revised page?
I've never been a keyword writer, don't pay attention to them at all other than in the title. That's not to say that all the key words dissapear, you can't really avoid using "reactor core" on a page about nuclear power, but you can rewrite:
In 1987, Sue Jones was the first woman in the U.S. licensed to supervise loading new fuel rods into a reactor core.
Reloading reactor cores with fresh nuclear fuel was the sole province of men until Mrs. S. Jones was certified on 6/5/87.
Forgive the example, I don't know beans about the industry, but when you do know what you're talking about, you can always say the same thing differently. My focus in a rewrite is to deliver the same information to my readers, otherwise it would be misinformation. And over the course of a two or three thousand word page, my guess is that if you counted keywords before and after, they wouldn't differ all that much.
I don't know why you're suggesting full Panda recovery is a myth, I did some research and found around a better than 10% full recover rate from the February Panda, wouldn't have bothered to start experimenting with my sites otherwsie.
I have no idea what my pre and post Panda rankings are for particular key words and phrases, other than some super popular phrases that caught me eye. I do know that on my site that recovered fully from Panda on its own and then started getting over 3X as much Google traffic on the following update, the phrases as reported by Analytics had shifted in many cases, and the content hadn't been touched.
In any case, I started doing these experiments to find out what's going on. You can argue with my conclusions - I can argue with my conclusions as well, but there's an experimental basis, I'm not just reporting what I feel.