[Patch] libstdc++/16612

[Benjamin Kosnik]

>the below is what I have tested (with Benjamin's help) as a stop-gap fix
>for the applications outlined in PR 16612: indeed, we have tested it
>together with the soon-to-be-contributed shared memory allocator.

Thanks Paolo!

I guess I'm still not sure why this is conditional. Can you explain your
reasoning here? I don't see the _S_empty_rep checks as being necessary
for correctness or performance. What am I missing?

-benjamin