To your first question, you have indeed correctly grasped what I was suggesting and I can see no sound reason why it isnt so, but equally I dont have any example to show where it has been done and I suspect that I am on my own in suspecting that it may not only be valid but in certain ways better. Much of the intricacy of best practice heat treatment involves methods/times/temperatures/protocols to minimize the deleterious effects that otherwise accrue like excessive grain growth, grain boundary precipitation, diffusion in, diffusion out, carbide clumping and so on. Clearly volumes have been written and lives spent on such intricacies and not without cause. Fabulous results can be achieved that way, but I cant help wondering if there isnt a Gordian Knot situation going on for this very specific case.
Industrially I can see issues but when a number of skilled people come forwards and say they have given it their best shot, aiming for strains of 2.5 to 3.5, tested the results and come to nothing but grief, it is just too hard to do and doesnt produce the results, I guess I will just shrug and consign the idea to the pile of things that sound like they should work but dont. Works for the steel belts in your tires, though.
As to the second, Mete is quite right in pointing out that it is very, very complicated to say exactly what is going to happen in a specific case in advance. Tatsuo Inoue, co-author of the Handbook of Residual Stress and Deformation of Steel, did some detailed work specific to the process as used in a traditional Japanese katana. The version of this presentation that I have found with the clearest illustrations is here:
http://www.shibuiswords.com/tatsuoinoue.htm
Section 7 relates and shows how complicated the calculations are, it all depends on the rates of cooling, the hardening response of the particular steel, the response of the quenching medium, the properties and application of the clay. As the edge cools fastest, thinnest clay and cross-section, it initially shrinks fastest and may (or may not) yield in tension, being restrained by a larger cross-section of the spine, and take a permanent stretch making it longer at thermal equilibrium, all else being equal. That is just residual stress or distortion. In this case you could be left with a belly out curve and an edge in residual compression as the shorter spine tries to restrain the now longer edge.
But all else is seldom equal, the cooler edge will at any given time point also be stronger than the hotter spine, so there will be complicated ratio of areas at different temperatures to consider. And most significantly martensite is for all the reasons given by Mete and bladsmth above less dense than pearlite so a pearlitic spine will again be shorter at room temperature than a martensitic edge and will attempt to compress it, resulting or exacerbating this curvature. But since the martensitic (or even just cooler) edge is stronger/harder than a pearlitic (or hotter) spine, for greater ratios of edge to spine, you could get the opposite on the temperature drop where the edge is strong enough to yield the spine in tension and induce an opposite residual distortion and leave you with a belly in curve at room temperature as noted. Or if the spine eventually transforms to all or mostly martensite despite cooling slower because you are using a high hardenability steel then you could again easily end up with a re-curve type curvature and an edge in residual tension.
To really know in advance of trying it, you would probably need to do the same sort of numerical analysis that Inoue does. But after you have the final result of a trial in your hand, you could measure the curvature, see where the hamon lies, knowing the microstructure and dimensions of the different zones and Youngs modulus, you could get a qualitative feel for what residual stresses you have induced in direction and rough magnitude.
Unrelated, but I thought the most interesting part of that presentation was the effect the clay had on the water quenching, reducing the steam blanketing from film boiling at the highest temperatures thus increasing the initial heat transfer when the material is best able to tolerate and most needs it and then decreasing the final heat transfer rate when you are below the pearlite nose and neither want nor can tolerate the much increased rate of heat transfer in the nucleate boiling regime.
Oops, I see while I was writing bladsmth has dealt with the guts of the last bit, but it is too late here to be bothered editing, so forgive the repetitive parts.
