I can't blame V8 engineers for not wanting to regress SunSpider. Sad or not—and, let me tell you, nobody who works on browsers likes this state of affairs when it comes to that benchmark—doing well on the popular benchmarks has to be an engineering goal in the environment browsers find themselves in.
FWIW xorshift1024* would almost certainly be faster than MWC1616 and it has no systematic failures on BigCrush. The only downside is that it uses more memory, but I think we can afford it. If not xorshift64* is also faster than MWC1616 and miles ahead of it in quality.