Galling, or at least localized metal removal, is something I have seen from time to time, and I had suspected that it was due more to corrosion than to the surfaces sticking to each other.

I could be wrong about that though, since an aluminum piston that seizes inside of an engine's ferrous cylinder (albeit at high temperature) does leave streaks of aluminum on the cylinder wall that must be dissolved with hydrochloric acid to restore the smooth iron surface.

Certainly grease might reduce corrosive interaction between the spindle and crankarm, and thus might be why Shimano did commence putting the grease in their final run of XT cranks.
Or perhaps it was because crank bolts are so often under-torqued by less-skilled installers, and grease would allow more of an interference fit to develop with whatever level of torque was applied. I suspect the latter.

I think it takes a lot of movement to cause galling between such dissimilar metals, mainly because the aluminum very quickly develops a thin film of oxide that acts a lot less like metal. But where greater motion and thus relative surface speed from the twisting of the spindle's ends occurs (as when mountain biking), that oxide film might have proven to be insufficient, though in my experience I have yet to have seen any examples of where apparent galling caused an actual problem with the fit on the spindle. At the point where bolts bottomed out on the end of the spindle, a bit of grinding work on the end of the spindle restored the full effect of the bolt tension.