Comment 8 for bug 1652323

The fix [1] for this bug https://bugs.launchpad.net/swift/+bug/1683689 introduced object server and diskfile changes that helps us work toward solution 1 enumerated in comment #5:

- we can now open expired diskfiles
- the x-backend-replication header causes x-delete-at older than 'now' to be allowed on a PUT

so we should be able to ssync an expired object (ssync sender can open it, receiver will accept it)

For EC case we need the additional step of being able to GET expired fragments. That could be achieved by including an x-backend-replication header with recosntructor fragment GET requests and modifying the object server to allow GET of expired object when the header is present (much the same as the change [1] that was made to the object server POST method)

[1] https://review.openstack.org/#/c/456921/