OK, on the axis between the back and front it is! I have come around to this view myself. Thanks for the keen eye, Elvis!
Absolutely. This was my main motive, actually, and I just took the opportunity to make some other small improvements along the way based on seeing the physical prototype in my hands. The amount of shrinkage in the indents seems to be about 100 microns (at least when measured along the short end of the indents), so I actually increased the space on each side by 100 microns (200 microns per indent, or 100 on each side of the aluminum part). That should give a little space for overflow epoxy and just a general safety margin. There is also a little shrinkage due to silicone molding (which increases with every casting), so the extra margin should help there too.
My model was getting messy and cumbersome to handle due to my having experimented so much and made a mess of the parametric history (I was learning as I was going with this project), so I decided to remake the model from scratch. I'm very happy to have done this, but in doing so, something changed in the model that only made it possible to create a shell that was 500 microns thick rather than the 1mm I used before. This is technically within spec for the detail acrylic printing process, but I'm a little concerned that the walls will be too flexible at this thickness to cast. We'll see. As a backup, I also ordered a solid print in nylon (which, interestingly, was still less expensive even though only a tiny fraction of the material compared to the shell version is used).
Nonsense. You're only to blame for having kept me on task on this project. And in any case, as I mentioned you you (@Eric_Ardros) in an email, I took the lack of fillets on your styrene masters as an excuse to experiment with 3D printing to see if it would serve our needs. The risk here on this project was low because I always knew that if the 3D prints didn't work out, I could always cast your styrene masters in resin and round over the edges by hand with sandpaper and then make a second mold from that. So it was the perfect project to experiment and fail a little.
Incidentally, I think the whole waste-mold-and-remold approach is something I may still take on this one. My plan was to gloss coat with a black urethane and then polish that, which is definitely a great approach for a project like this in general (Volpin used it to great effect with his translucent Curse Gaming awards). However, I'm still new to urethanes, so I'm a little concerned about 1) screwing up the master with fisheyes or something and then having to get a second non-trivially-expensive 3D print, and 2) even with masking, the paint may seep into the edges of the recesses for the metal parts and throw off the aforementioned carefully-calibrated tolerances. So I'll probably mold the 3D print, cast it in Smooth-Cast Onyx, and sand and polish that, and then make a final "production" mold based on that.
In short, I can find a clever way of making almost anything complicated.