This is turning into a very interesting thread.
The Laws of Physics definitely predict the metal will expand when heated. In most simple geometries the diameter of a hole will expand, as will the diameter of a shaft and so on. The geometry of a hole in a shape as complex as an aluminum cylinder head will change according to the Laws of Physics applied to the entire volume of the head, taking its internal and external features and shapes into consideration. It is not clear that the diameter will expand evenly or consistently for the entire length of the hole though.
In the end the hole for the spark plug will change dimensions, but before I would bet a lot of anything on the outcome, I would either check it with a guage hot and cold, or I would run a finite element analysis to identify how the shape changes as it heats up. Any relief you get from the diameter growing will likely be offset by the length change, making the threads a slightly different pitch, kind of like an interference fit. In addition, the mechanical properties of aluminum degrade with temperature, and a head at normal operating temperature is more likely to let go of the threads than it would be at room temperature.
The primary argument to change the plug cold is the engine was designed to be assembled at room temp and should be disassembled at the same temp. Any tricks of the trade, like adding penetrant when it is hot to help draw it into the thread are great to know, so thanks for the tips. Jim
1986 Euro 190E 2.3-16 (291,000 miles),
1998 E300D TurboDiesel, 231,000 miles -purchased with 45,000,
1988 300E 5-speed 252,000 miles,
1983 240D 4-speed, purchased w/136,000, now with 222,000 miles.
2009 ML320CDI Bluetec, 89,000 miles
1971 220D (250,000 miles plus, sold to father-in-law),
1975 240D (245,000 miles - died of body rot),
1991 350SD (176,560 miles, weakest Benz I have owned),
1999 C230 Sport (45,400 miles),
1982 240D (321,000 miles, put to sleep)