stop using ioends for direct write completions

[Christoph Hellwig]

On Thu, Jan 28, 2016 at 01:58:53PM -0800, Darrick J. Wong wrote:
> Aw, snap, I knew I'd forgotten something.  Yep, we'll need that... I think
> xfs_end_io_direct_write will have to sniff out the error status from "size"
> and either remap or discard the CoW allocations as appropriate.

I'd rather fix the direct I/O code to give us that information directly
(pun intended).  I'll add that to my short term todo list as it seems
useful for the existing code as well.