LU-10189 osd: handle PFID EA in LMA properly

LU-10189 osd: handle PFID EA in LMA properly

Originally, the issue was caused by old ldiskfs OST device
with 256-bytes sized inode. Because the inode inline space
was very limited, we have to store the PFID EA inside LMA
EA for stripe and PFL component information.

When we restore the OST from such old OST via server side
file level backup, then such composite LMA will be on the
new OST even if the new OST inode has enough inline space
to hold separated PFID EA.

In futher, if we migrate the old OST from ldiskfs to ZFS,
then such composite LMA will also be on the ZFS based OST
although the PFID EA can be stroed independently on ZFS.

So the OSD logic, in spite of ldiskfs or ZFS, needs to
understand the composite LMA EA, and handle it properly.

Signed-off-by: Fan Yong <fan.yong@intel.com>
Change-Id: I2b66787e725e13da7984f1bc2df45760dfbe4c4d
Reviewed-on: https://review.whamcloud.com/29696
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-by: Lai Siyao <lai.siyao@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>