It's too weird to have a meta tag that is only useful for websites with a white background. The narrow use case for this was solved years ago with the hcard micro format. We should be pushing microdata, not creating new niche standards.
You could even consolidate this so you had rel="logo-favicon" and the favicon just become another aspect of your logo.
All that said, I think it's a solution looking for... well, not a "problem" really, because the problem is obvious. Rather, the whole thing seems like a solution to a problem that not enough people have. I'd wager the 2011 proposal would have gotten more traction and some format adopted if this were really a pressing concern.
You know what? What I'd like to see is some kind of de facto standard rel="foo" database that, a bit like the WHATWG, kind of becomes a place where you can submit rel types, spec a schema behind it, go through some semi-formal vetting and standardization process, and be able to crawl this information in a semantically-meaningful way (in addition to just researching the different types). It'd also be nice to formalize a migration path from a <link rel="x-foo"> as a prototype declaration to a <link rel="foo"> as a 'standard'.
The closest we have to this is probably the HTML spec and WHATWG's page (http://blog.whatwg.org/the-road-to-html-5-link-relations).
That said, for this case, this right click implementation seems smart.