ocfs2: __ocfs2_abort() should not enable panic for local mounts

In a clustered setup, we have to panic the box on journal abort. This is
because we don't have the facility to go hard readonly. With hard ro, another
node would detect node failure and initiate recovery.

Having said that, we shouldn't force panic if the volume is mounted locally.
This patch defers the handling to the mount option, errors.

Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>