To sort out the lead angle problem, the first thing you need is the lead angle. Something like 5 degrees I think (longhand mathematics, verify that), which means that the leading edge of your threading tool, to RUB the work without (much) interference, would have to match that angle. Plus you want a working clearance. So that leading edge needs a clearance angle of MORE than 5 degrees.
The shims, or otherwise rotating the cutting tool do just that, they rotate the tool. So if you did not have enough clearance, rotating the tool "towards" the cut will effectively add more clearance to the leading edge. And also reduce (or reverse) any rake angle on the top of the tool.
Yeah, that's the idea. I'm guessing it leans more towards insert tools, as they're harder to change such angles on, and/or require different cutters or whatever else. With HSS tools, it's literally just the clearance on the leading edge that gets in the way. Go to the grinder and knock that back (if it interferes), and you're done. Otherwise, or if that's impractical for whatever reason (including insert tooling that doesn't lend well to that), then you'd rotate the tool so that the leading face does not rub behind the cutting edge.
If you have a sample of known good to mediocre, but not "shot" acme rod, just by hand, stick the cutting tool into the thread. Straight, level, and eyeballed on center. If the cutter drops in well, you're good. If the cutter wants to "lean left" as you insert it, there's not enough clearance. That little science project in it's self might be the best explanation of what's going on that you could find...
Also, mmcmdl is onto something there. If you run the thread in with a "normal" threading tool, right to depth (with or without using the compound, whatever the lathe and/or the tool likes best), it won't make a full width thread. But it will "take out" the center of the cut you want. That makes life a lot better for the Acme cutter, which by the time 5 TPI comes around, is a pretty significant form tool. With the "middle" of the cut out of the picture, the two side cuts are collectively a lot easier. All the way down to the "point" of the "regular thread" cut. Just the separation of the left and right sides, the two simultaneous cuts are far lighter than when the whole chip was conjoined as one.