For those of us who forge our blades, and I've forged hundreds of them over the last 20+ years, normalization is a key step to de-stressing the blade prior to heat treat. Often I'll normalize 2-3 times just to make sure my blades are nice and "relaxed" before I start the heat treating process.
What you bring up in the first and second sentence is a good point for a person to consider when forging a blade if you want to economize your efforts. Mash metal in one direction, and it will spread in every other direction. Different shapes of hammer faces can be used to minimize or maximize that propensity so that the 'smith controls the shaping process more accurately. A good example would be the difference in metal movement when you compare a flat or rounded hammer face striking straight down versus a straight pein hammer doing the same. All other things being equal, the flat or round faced hammer will spread the metal in every direction as it compresses the metal between the hammer face and the anvil. A straight pein, however, will minimize the spread parallel to the long axis of the hammer and maximize the spread perpendicular to the hammer blow as the metal is compressed between the hammer face and the anvil.
The effect of metal movement away from the hammer face is shown graphically in the knifemaking tutorial I wrote and linked to in a previous post, and again in the belt axe tutorial I did on the same site (link here: http://paleoplanet69529.yuku.com/top...e#.VpaNMI-cEdV
If a blade is very asymmetrical, then yes, some warpage will occur, however, with only slight asymmetry, it's negligible. Interestingly enough, after the quench, if you leave the blade a bit hot, you'll have about 10 seconds to straighten the blade before it takes a "set" and is fully hardened.
I always got a kick out of showing people a trick when I was quenching hardenable steel for demonstration purposes--I'd pull it out, still smoking, immediately put a big bend it in, wait about 20 seconds, then shatter it with a hammer.
Japanese 'smiths water quench, and the tamagahane (the traditional steel used for knife making) they select for use out of the bloom produced by the tatara (a small Japanese style smelting furnace) is typically about 60 points carbon--equivalent to our 1060 low alloy steel. The upwards curvature is mostly caused by differential cooling rates between the edge and the spine, where the spine and a portion of the sides of the blade are coated in clay slip mixture, dried, then the blade heated to the critical temperature and water-quenched. As you mention, there is a fairly significant loss rate during that process. An interesting side effect to the clay coating during the heat treat process is the formation of a hamon, which visually delineates the transition between martensitic and pearlitic steel in the blade. Nioi are the speckles of martensite surrounded by pearlite in the transition zone. All that aside, my interest lies more in making mokume gane (literally "wood-grained metal"), which is a forge-welded, multi-layered billet of dissimilar non-ferrous metals.