The design and drawing are what was intended to be produced while the actual production frames may vary, especially on hand made models. I normally assign frame angles by the design/build intention, since measuring 10 hand built frames down to the tenth of a degree will likely result in 10 slightly different results, especially if they are different sizes. I think there is way too much over-analysis going on here.

Using one physical example of a specific model to measure is also a problem, I'll see if I can measure at least 10 to maybe 20 Schwinn fillet-brazed frames to find the angles based off the top tube. That won't happen today though.