Apparently the coefficient of thermal expansion for 6XXX and 7XXX alloys is around .000025 (or so) per degree Celsius. Thus a fuel tank that is 2 meters long will grow about 5 cm if heated 100 degrees C above ambient.
Our tanks will likely never see a delta temp change of more than 50 c, and are less than 2 meters long, so the max change would realistically be less than 2.5 cm. At normal ambient, that means it could grow by 1.25 cm and shrink by 1.25 cm at the extremes. Thus, one could rigidly mount one end and engineer the other end to accommodate a swing of slightly less than 1.25 cm in each direction.
1.25 cm is just under a half inch (finally we get to good ole 'Merican numbers here), so mounting the end that moves centered in a half-inch slot would allow the tank to move with no stresses. Using foam, HPDM or similar material would go a long way towards stopping rattles, I expect.
So, that's what I'm going to do.
-Will
PS: I haven't stayed in a Holiday Inn Express for some time now, so if anyone wants to correct my math, or analysis, please do!