Flux: most of the shunts I deal with have the sense-element silver-soldered into the brass block. . . . so if you use a low-melt solder to assemble the multiple shunts, it's not a problem. Other brands may vary?
I agree with the opposing corners for power and sense argument; I'm not sure that the position of the low-value shunt matters IF you get good contact. . . but in a marginal contact situation, you're probably right that it would be best in the center.
Dean: there's a lot of folklore in the high-current connection fraternity. the one thing all the anectdotes seem to agree on is the libral use of silver plating. . . . and the avoidance of zinc plated hardware. there seems to be some voodoo associated with threaded hardware. (some use silver paint on the threads, a few, lead antisieze, most use silver goop or dry threads.)
All I can relate is personal experience, which says stainless hardware in copper busses lead to expensive failures.
I suspect that our folklores are different because of the environmental differences: industry vs environmentally controlled science lab. (the lab is a nicer building inside than my office. . .) where I am, we optimise for conductivity and don't worry overmuch about environmental factors. For a shunt or a magnet that's riding the ragged edge of thermal feasability, the extra resistance of the oxide layer in the stainless creates higher resistance and higher interface temperatures. . . =more stress on the part and skewed accuracy. plus, with the amount of torque I'd be putting on a stainless bolts to assemble the shunts with good contact pressures, in my experience, the stainless bolts tend to gall their threads and jam, leading to uneven contact pressures and sometimes even problems getting the thing apart. even with that, stainless is undoubtably a better choice than plated carbon steel . . . the zinc corrodes too easily and iron oxide doesn't conduct worth a damn.
we use the same connection methodology (bronze hardware, silver plating, silver mesh) for connections with high currents (30Ka+) and ultra-fast risetimes(~1/2 ns.) again, it's a question of what you're optimising for; in our case conductivity in a controlled environment is the driving factor. That said, I've never had one of our lugs work loose with bronze and silver mesh. . . . but did have to make a new current bus for a magnet winding that had been wired with stainless. (no belville washers, just a split lockwasher, and tapped directly into the copper. it got loose and burned up the contact block. 10 tons of magnet isn't something you just toss in the mill to renew the mating surface on. . . so you could say, 'once burned, twice shy. . .')
for your applications,(motor drives, heat-treating ovens, etc) stainless would (obviously) be ok, maybe even recommended if you get a lot of odd process vapors. I'd be worried about differential thermal expansion coefficients, myself, but if you've had good reliablility doing it your way, then your solution is well matched to your application.
the belville washer trick is new to me (I like it on the face of it); maybe that's the secret to dealing with differing thermal co-efficients?
-Dan