The problem with that idea is that many parts of Android consists of fundamental changes to the Linux kernel. Google has repeatedly tried to get their changes incorporated, but Linus and the rest of the core linux developers has vetoed most of these changes.
The objections mostly deal with Android performing certain types of power optimizations that are detrimental to the mainline kernel.
So even if Google did "open" up android completely, it would represent a fork of the linux kernel instead of being "embraced".
The only alternative is to give the linux devs free rein to change Android and rebrand it (since you can't use the android moniker).
Getting code into the Linux kernel requires patience and a lot of work to conform to the standards set by the developers. I saw a great talk at ELC 2010 [1] from Greg Kroah-Hartman on the problems they faced when trying to get the Android changes into the kernel. He also did a short write-up of the problems[2].
The objections mostly deal with Android performing certain types of power optimizations that are detrimental to the mainline kernel.
So even if Google did "open" up android completely, it would represent a fork of the linux kernel instead of being "embraced".
The only alternative is to give the linux devs free rein to change Android and rebrand it (since you can't use the android moniker).