[Patch] Uglify some names in _Rb_tree

[Paolo Carlini]

Howard Hinnant wrote:

> That sounds good.  I would've suggested storing a pair<key, value> 
> (say two ints) in the set, and using a custom comparator that only 
> looked at first.

Your suggestion is *very* cute, much more than mine, no doubt about
that. Now I'll just start implementing the tests - I also want
consistence between multiset and multimap - and if my trivial approach
turns out to allow a sufficient coverage, I'll go with it, otherwise
will choose yours.

Thanks a lot for your quick help, anyway,
Paolo.