First patch to simplify <algorithm>

[Gabriel Dos Reis]

Paolo Carlini <pcarlini@suse.de> writes:

| >There might arise some issues with associated types, though they are
| >easily dealt with.
| >
| ... could you please clarify a bit this final note?

I thought our implementation was implicitly making the assumption that
some non-pointer type function objects derive from unary_function<> or
binary_function<> (therefore would provide xxx::result_type and
likes).  But, I just skimmed through and I can't find any such thing. 
(Somehow I got the idea that that were the case, while working on
something else). Sorry for the false alarm.

-- Gaby