you can call it a interference fit if you want. That's just semantics.
Well, it would certainly be semantics if the terms meant the same physical thing, but you can tell by the units they don't, so I'm not sure whether it can be laid to semantics or not. Maybe so, in the long term, as misused terms often become common over time. As you note, SAAMI hasn't made the jump yet.
The problem with the reassignment of technical terms is always the potential for confusion. We had a hydraulic technician on the board once who couldn't discern pressure from force because the industry lingo referred to hydraulic line ratings as being in "pounds." It is actually in pounds per square inch, of course, but they are in the habit of dropping the last three words as an abbreviation, and this fellow took it literally. It caused him to make some very odd ballistic calculations because he couldn't see getting pounds force by dividing pressure by area. If the pressure was 60,000 psi, he thought it applied 60,000 pounds of force to any amount of area. I had a heck of an argument going with him because he wouldn't believe it and kept posting wrong numbers.
At issue here is that interference fit has units in inches while tension is in psi, and the one can't be converted to the other. So the inteference fit is, as SAAMI says, more literally a measure of what you use to cause hoop tensile stress (tension) but isn't the tension itself. You could certainly say you were setting the tension with it.
Anyway, your experience with pulled brass matches what Ive always seen with resizing and expanding necks, which is that they only spring back about 0.001" or so (varies with hardness and caliber), showing the elastic limit is about that much. After that, you get less increase in force from each additional thousandth of stretch. Not zero, just a lot less, as you are then past the knee of the curve.
Here's stress v strain a curve from a government report.