VFS: clean ups and simplifications related to name lookup

Hi,

 while we wait for Al to have time to look at my previous large patch
 set, I wonder if there is any change these first 4 could be considered
 for the next merge window.  That would then allow me to work with other
 filesystem developers to get the simple cleanups done.
 
 I think these are all valuable in themselves even if the larger changes
 I want need significant revision.

 [PATCH 1/4] VFS: fix various typos in documentation for

   Simple documentation fixes - nothing new.

 [PATCH 2/4] VFS: enhance d_splice_alias() to handle in-lookup

   When I submitted this some months ago Al didn't like the fact that it
   transparently handled various sorts of dentries and wanted
   d_splice_alias_hashed() or similar.  After some consideration I think
   this concern doesn't match current VFS behaviour.

   As outlined the in the patch description there are various
   circumstances where an inode_operation might be called with one of a
   selection of dentry types.  For example ->atomic_open might be called
   with hashed-negative or in-lookup.  For an atomic_open handler to be
   able to work with this it is best served by an interface which will
   accept either, rather than being required to pick an interface after
   testing what it has been given.

 [PATCH 3/4] VFS: allow d_alloc_name() to be used with ->d_hash

   This is a minor improvement and will only allow one other uses to
   make use of d_alloc_name().  However I believe that improves
   uniformity and will ultimately lead to simpler clearer interfaces.

 [PATCH 4/4] VFS: use global wait-queue table for d_alloc_parallel()

   This simplifies an interface.

Thanks,
NeilBrown