[PATCH] Exploit move semantics in std::accumulate.

[Paolo Carlini]

> Indeed. Would it make sense to provide overloads of some selected algorithms, like std::accumulate, for ranges of string::iterator? For different reasons, we are already doing something similar for std::copy and std::fill vs deque::iterator, for example.

Nevermind. It's a somewhat different situation, as the performance testcase shows. Humm...

Paolo