[RFC] Removal of allocator::construct / destroy vs our containers

[Paolo Carlini]

Benjamin Kosnik wrote:

> Cool! I guess this waits for a more elaborate discussion in Toronto as 
> well...

Oh well, it's just a small tweak. Frankly, I should have done this 
already when I added the various _M_get_* everywhere (with Howard's help).

I'm thinking of applying the patch even earlier than Toronto, that way, 
in case construct / destroy will stay, we are ok, performance-wise, 
anyway. Otherwise, in case construct / destroy will actually go away, it 
would be only matter of tweaking a bit the posted patch, matter of a few 
lines. That way we could relax, not having to remember for months to 
apply this last patch in case construct / destroy eventually stay...

Paolo.