Yeah, I can understand how they could get overly tight and burn up, but it's strange that they don't have you re-tighten it to another lesser torque spec after the fact. It creeps me out.

Hasn't failed yet, I suppose.
Actually, you are tightening it to a known torque, you just are doing it backwards, which in this case is best.
When you put in new bearings and races, you seat them against the stops. ( I know, they both don't have stops, but its late)
However, its almost impossible to get them seated totally and evenly.
So, you assemble every thing and start to torque the nut.
As you approach 50 Ft-lbs, the parts take up the last of the slack and align themselves.
Tests have shown that by the time you reach 50, unless something major is wrong, you are steel on steel.
At this point, given the hardness of everything involved, going to 100 would not involve more than a part of a degree of turn.
If you were to graph the torque vs rotation of the nut, it would be a fairly straight line running at about a 45 degree angle, until all the slack was out of the parts.
Then the torque would pretty much go straight up with almost no rotation at all.
At this point, if you back off to slack and re-torque, some of the parts might flex and not be seated anymore. So the final torque would not be accurate.
Since they know the angle of the bearings, and the number of threads per inch of the nut, they can predict the final torque value from the amount of rotation of the nut.
A lot of the bigger, higher load bearings are done this way, because the bigger they are the more flex they have, and the higher the load the less forgiving they are.
I have done them both ways according to the manuals. Over the years, I have found more of the ones I loosened and re torqued to come up too loose after a few miles than the ones I torqued and backed off a certain amount.
NOTE: I got that from an old SAE book I read years ago. It even had the torque vs rotation graph in it.
Given that I read it bout the time your K5s were just rolling off the assembly line, I have no idea where to look that up now.
J.