I'm not quite understanding how profiles work. A profile is supposed to do a good job of mapping the gamut.? These profiles are generated from measurements of prints of grayscales and color charts.
Why, when proofing, are extra steps required to tweak images for the paper to almost try to get it back to what it looked like before selecting the proof option?
As I understand it, the reason extra steps are required when proofing is this. The effect of printing is to distort the image you see on screen (which itself may or may not be a "true" representation of the pixels lying in the image). The distortion arises because of the interaction between printer, ink and paper.
Soft proofing allows you to see on screen more or less what the effect of that distortion will be on your image. You can then apply other distortions in advance of printing, in the hope that the two sets of distortion will cancel each other out.
I can't even begin to address your other points and I'm sure if I what I've written above is wrong, it'll be corrected. There are some clever people on this forum!