Not really an answer, more like musings on the subject (though it does answer your side question about VS2010).
The problem with C++ refactoring is that, unlike code completion which only needs to be "good enough", refactoring operations such as "rename method" should just work. If you have to manually do a Find on your code afterwards to verify that it didn't skip any methods, what's the point of the feature?
This means that C++ editor with refactoring support must have perfect understanding of C++ syntax and all compile-time semantics. This becomes important when templates get involved - a method call that has to be renamed might be made on a receiver which is a result of some nontrivial template function call (consider the case where it's std::bind
!), or a combination of such. This means that supporting C++ refactoring is a task that is an order of magnitude more difficult - you have to lift all template instantiation logic from the compiler, including template specialization, SFINAE, etc. Keeping in mind that templates are Turing-complete, you effectively end up writing a full-fledged interpreter.
Getting back to real world. Current and past versions of Visual Studio used a simplified parser for all editor-related tasks, which is why IntelliSense worked sporadically, failing on many less trivial expressions, and why there is no refactoring (it just wouldn't work reliably).
In VS2010, this has been replaced with a new engine which is based on the front-end to EDG C++ compiler - as such, it can fully grok arbitrary complex C++ code. I've actually tested it on some very non-trivial things and blogged about the results, which are pretty amazing.
Unfortunately, the team (apparently) didn't have time to fully use the capabilities new parser, so currently they're only used for IntelliSense and Find References/Symbols. Thus, there's no out-of-the-box C++ refactoring in VS2010 beta.
The good part of it is that now the new parser is there, reliable refactoring is possible to implement, and I see very few excuses not to use the data to enable full-fledged C++ refactoring eventually. I do not know if there are any specific plans for this, and I haven't seen any public announcements, but I'd be surprised if this feature isn't at least considered for the next release. Your best bet to affect it is to open a Microsoft Connect feature request for C++ refactoring (surprisingly, I cannot find an existing one... surely there are many people who want this?) and vote for it.