[patch, rfc] Fortran async I/O support

[Dominique d'Humières]

Hi Nicolas,

I have been experimenting with your patch on top of revision r260849 and I got some regressions at run time:

I see the speed up for the test in pr25829 comment 41, but only with -O0. If I compile the test with -O1 or above, I get a segfault at run time.

I also see segfaults at run time for some tests in PRs 29951, 30014, 34876, 35699, 35840, 37083, and a variant of 51334.

I have now reverted the patch, but I can reapply it if you need more information.

Thanks for working on this.

Dominique