As I understand it, this allows the researchers to decrypt the microcode but not to sign their own modified microcode, is that correct?

Given a magic wand, I would update the link URL to not take one directly to the comments section

It's somewhat understandable; on Ars, you can go to comments view and expand the full story, then forget you're actually on the comments.

Ooops, my bad. Sadly, I can only update the title.

I seem to recall hearing that some CPUs were hobbled or otherwise limited by software for product diferentiation, but were otherwise identical.

I wonder if this can be used to unlock additional cores or features on some CPUs.

they are not necessarily identical. however during fabrication, there might be deffects in some cores, so Intel can disable some cores and market it as a lower end product. it is called binning

binning locking down multiplier change? lets not kid ourselves