I just had to pick this out, because it seems to me you're overlooking another method- what you might call 2.5- where you don't just repackage old material, but you expand and add to it without invalidating the old stuff (as in your case 3).
To stick to Greyhawk as an example, there are hundreds of things or more that are only sketched out. Most of the cities have only the barest descriptions- many are only names and populations. There are countries that have barely been scratched, political relationships that have never been explored, tons and tons of things that can be easily expanded without "overwriting" what has come before (which usually alienates the fans).
And that sounds good to me. In my ideal world, we'd see tons of setting stuff - old and new. But for me, if its the choice of seeing an old classic like Greyhawk or FR re-hashed (and deepened, as you say), or an entirely new world developed, I'd like to go with the latter.
The simple problem with this: they have too many worlds, rebooted too many times.
The last thing they should do is come up with yet another one. Whats the point? They don't need it for new players, everything is new to them. And for old players, they are competing with all those existing worlds, which many of those older players prefer. Its very, very hard to justify.
There's a slight contradiction here, Dave. "too many worlds, rebooted too many times" yet you don't want them to come up with a new world, but instead reboot one of the ones they've already "rebooted too many times?"
In the end, though, it probably matters less
which world they support and more
how they support it. A very well supported and published Greyhawk or FR will be much better than a crappy new setting. I think part of 4e's relative failure was the lack of a living, developing world, which prevented WotC from saying, "Look, this is how D&D 4e feels."
A living setting gives an edition (or game) a center, a heart.