I've been back at work on this. Waiting for the relay, I got impatient. I ended up doing pretty much all of the electrolytic caps through the whole thing, without any step by step testing

. I was, however, much more diligent about taking pictures, going one part at a time, and triple checking everything. The main power amp board is the only thing I haven't yet touched, and I will definitely test it out before starting on that.
First up, we have the main power supply caps.
For these, I restuffed the cans using 2x 6800uh 80V each (nichicon low-ish ESR. not sure what series...). I measured ahead to make sure they would fit. Since they wouldn't fit side by side like in Velox's post, I taped them together top to top, wired a positive to a negative, and extended the leads. See Pic 1. The basic setup, as you can see in the pic, is that I attached a lead (blue) to the negative side of cap 1, then ran the white wire from the negative side of cap
2 over to the positive side of cap 1 and left a tail there so I can wire that to ground. Then on the positive side of cap 2, I attached the red wire and ran that down the other side so the whole shebang can be packed into the old cans and all 3 contacts can come out the bottom. I'm pretty sure both of them were done this way, but not 100% sure. MISTAKE.

I'm not risking it. Gonna pull it out and double check the polarities before I send any power to this thing.
To cut open the cans, I used a hacksaw. The first one was rough. I didn't realize that you don't have to cut the thing in half straight through... you can just cut through the aluminum can around the outside. For the second, I got a much neater cut. I used some scrap wood to whip up a makeshift wedge/block to hold the can in a miter box, and I rotated the can as I went, so my blade never went more than a few mm in. Just enough to cut through the casing. MUCH easier. Anyway, see pic 2 for the beginning of the surgery.
I used a 14in pipe wrench to twist the guts out of the cans. Worked very well. Almost all of it came out in a big chunk and I just had to use a screwdriver to remove a small amount of remaining waxy coating from the inside of the top of each can.I drilled a few holes in the phenolic plate at the bottom of the cans for the wires to come out of. Tip for folks out there: do this outside or at least in a bathroom with the vent fan on. When you cut/break phenolic resin, it releases gasses that smell terrible and are terrible for you. If I had it to do over again, I wouldn't have broken the original tabs off the bottoms of the cans. I would have left at least the "A" and "B" tabs in place to use as solder lugs for the new wires. Pics 3 and 4 show various stages of the restuffing work. I had some 1/2" open cell foam packing material that came with something or other. This stuff worked really perfectly to line the cans and hold the new caps in place. Once the new guts were installed, I used black duct tape to stick the cans back together. The clamps help hold the two halves in place, as well. See pics 5 through 7.
Again, I wish I had looked back at Velox's post here and thought to leave at least the two tabs for the
non-ground wires (red and blue in my reconstruction). That might have made me be more careful to double check the polarity of my reconstruction. I also wish I had planned to color match the positive and negative "terminal" wires coming out of the restuffed cans with the red and white wires inside the amp. Lessons for next time.
I'll post a pic of the final installation after I verify the polarity of the restuffed cans.
While I was working on this, my x-tronic 3020 soldering iron bit the dust. I took it apart but couldn't identify the problem. it isn't the switch, or a blown fuse, and there is no visible scorching or blown components. It's only 9 months old, and although it has a 3 year warranty, I got it on Amazon so I'm not sure how likely I am to get a replacement. (still gonna try, though)
Anyway, rather than get another potentially short-lived iron, I decided to go with something that is sold locally and a bit more serious. I checked out Dayton Audio/Parts Express and ended up getting the Stahl DSDS Soldering Station. It has an iron and a desoldering gun, both with digital temperature control. The reviews were positive and it has a year warranty so I took the $175 risk figuring that with the dealer being local-ish (35 minutes away), I would at least have an easier time of it if there ends up being a warranty claim. I also wasn't prepared to shell out $100ish for a hakko iron, and another $400ish for the hakko desoldering gun.
Initial impressions are that this is a good piece of equipment. The desoldering gun works pretty well in my first 5 or 6 tests. Like the desoldering pumps I have used, you
do have to clean the suction tube out every couple of uses, but the station comes with 3 little poker tools that makes it pretty easy. Certainly easier than disassembling the pencil style hand pump I've used for the last year. The suction is good, and the tip of the gun heats up reasonably quickly. The iron itself heats up fast and seems to maintain temp reasonably well. It comes with just the standard pencil type tip for the iron, so I wish I had ordered the $7 set of interchangeable tips to go with it. I've grown quite fond of using a small, flat wedge shaped tip. That shape seemed to hold and transfer heat better than the standard pencil style tip on my x-tronic. I suspect that a similar tip for this Stahl iron will have comparable results. The only knock I have on the setup is that there's no solder spool holder. It took a bit of work, and minor disassembly of the new Stahl DSDS, but it wasn't too difficult to just attach the one from the x-tronic to the side of the DSDS.
Next steps for the 2330B are to double check the polarity in the one can, put the new transistor and relay on the power supply board, and give it a listen.
If all is well, I'll finish off the main power amp board and button it all up. If all is NOT well, you can be sure I'll be back here either documenting the issue and fix, or soliciting input from the AK braintrust.