Plan 'A' Seems to Have Worked.

How often does that happen, anyway?

Over the last three days I've managed to build an image of the failed raid partition in its active state. (Imaging the failed individual drives happened earlier as a precaution, but I wasn't looking forward to raiding the images to get something that looked like a filesystem...).

In any case, the image appears to be a real filesystem. I can mount it and everything, although I can't do a fsck because reiser needs to be able to write to a partition to do a 'read-only' fsck. WTF?  In any case, we'll be abandoning reiser in the very near future anyway. I have already partitioned a new drive to mirror the partitions in the old array and as soon as I can clear enough space to physically open up the webserver and swap drives, I'll be able to start dd-ing the images into their respective partitions and then seeing what fsck says about their consistency.

So, barring a catastrophe, it looks good to be up and running again by the weekend.