Let me see if I understand. You define capture sharpening as global sharpening intended to deal with the effects of anti-aliasing which is by definition "low level" and "not aggressive" and something that the camera itself can do pretty well most of the time. You distinguish it from selective sharpening, which is done as an adjustment step, typically with USM. This is a useful distinction. I had previously considered both to be "capture sharpening."
I have found that I can always do a better job of sharpening than my camera -- at least one that is more to my taste . I therefore always zero out camera sharpening and replace it with a USM adjustment step. I think you would call such a step "capture sharpening." If I need some selective sharpening, I add one or more USM or high-pass adjustment steps, each involving some kind of selection.
You say the the primary purpose of USM is to control noise. While I agree that it can be used for this, I don't think that is its primary purpose. I believe its primary purpose is to enhance detail -- either low frequency detail (like contours) or high frequency detail (like surface texture). Within a given scene, you may have areas of both kinds. This calls for selective application of USM in separate adjustment steps, one for each area, each with different intensity and radius.