Great overview video and it's wonderful to see the puzzle progressing so much. I'm a big fan of your BubbloidXYZ series so I hope they keep coming!
I'm glad the hot water trick worked. I discovered that trick while cleaning pieces in hot water after tumbling. I'd test the tolerance and notice they'd fit together much easier than later when I'd be assembling the puzzle cold. There have been a few puzzles that I have assembled literately under water. I've filled my sink up with water at around 50C and put together some of the really hard pieces fully submerged. I'm not sure how much the water acts like a lubricant when WSF isn't polished but it's an incredibly good lubricant once the pieces are tumbled. Some of puzzles I've polished have turned better wet than dry and lubricated with silicone oil.
I didn't know you'd made the external walls so thick. It's a great idea though. It'll make the puzzle feel much more solid and it'll also make the tumbling process much easier on me.
Finally, in watching your video, I noticed a really common cosmetic print flaw in some of the pieces where the faces are slightly sunken in:
bb123_print_flaw.png [ 120.73 KiB | Viewed 877 times ]
I've seen this flaw and a few flaws very similar to it many times on many different puzzle prints. I believe it happens when the face of a piece is parallel to the printing layers (the face is normal to the gravity vector). I suspect the issue has to do with the expansion of nylon due to heat in the printer causing the face (or maybe the walls bordering the face) to be slightly offset.
I believe the best way to prevent this issue is to apply a small rotation (maybe 20 degrees?) to the pieces when you export them so their main external face(s) won't be parallel to the printing plane. TomZ's program will pack them just fine even with this rotation. Perhaps some designers (Oskar?) have tested the affect of piece orientation on cosmetics?