cleanup truncate handling in ecryptfs v3

Hi Tyler,

this series cleans up the truncate handling in ecryptfs.  I did
it as preparation for some changes into size changing truncate
VFS interfaces I'm looking into in the moment.  The changes have
passed the regression test suite in the userspace ecryptfs
repository and against the ecryptfs next branch.

Changes since v2:
 - change the calling convention to only pass the upper ia to
   __ecryptfs_truncate

Changes since v1:
 - fix a pre-existing spelling error in a comment
 - pass an explicit new_size argument to the main truncate routines
 - fix a comment that kept obsolete information

Diffstat:
 inode.c |  257 ++++++++++++++++++++++++++++------------------------------------
 1 file changed, 115 insertions(+), 142 deletions(-)