Re 1) I'm not complaining -- but instead advocating -- glibc in this context. Since glibc has a genesis that is fundamentally different from that of Bionic, you can't infer anything from the fact that glibc is legally safe according to industry understanding. We wouldn't be having this discussion if Bionic used glibc, or a similar library derived from the sanitized headers.
It seems @vonbrand that you don't fully understand the genesis of the sanitized headers, so please educate yourself before you confuse people here. Again, I'm all for glibc, but Bionic is a very different animal.
Re 2) What Linus Torvalds and RMS said is not related to Bionic and doesn't answer any of the legal questions at issue here.
Re 3.1) You conflate Google's objective with the legality of its chosen means in a way that distorts what I said. The objective clearly was to work around, or "circumvent", the GPL. TheFreeDictionary defines "to circumvent" in this context as "to go around; bypass; avoid or get around by artful maneuvering." So it can be legal or illegal. In that nonjudgmental sense, there's no doubt Google sought to circumvent the GPL, and the use of a script to cut out parts of approximately 750 header files to declare the remainder (which is actually the essence of those files) non-copyrightable is at least "artful".
Re 3.2) Your bet just shows that you're unlike the decision makers at commercial software companies (or hardware and other companies that build software for commercial purposes). His bet is meaningless -- and so is your support of it -- because real decision makers will take a different perspective and Google will only convince them by actually delivering a fix, not by saying it's easy to deliver one if and when needed.