Reconstructor once mode doesn't aggregate reconstruction_last/time

When we run the object-reconstructor with once and multiple workers mode (e.g. reconstructor_workers = 2) like "swift-object-reconstructor /etc/swift/object-server/1.conf.d --once", the reconstructor reports per_disk stat like:

{"object_reconstruction_per_disk": {"sdb1": {"object_reconstruction_last": 1503453604.363095, "object_reconstruction_time": 7.956425348917643e-05, "pid": 6112}, "sdb5": {"object_reconstruction_last": 1503453604.374259, "object_reconstruction_time": 0.00025614897410074867, "pid": 6111}}}

However, we cannot find the aggregation stat at the root node in the dict that will be created by parent process periodically when we run the reconstructor as *FOREVER* mode.

And then, as told as above, that lost of aggregation happens "once" mode so that it won't be created forever.

I think the possible solution is likely:

diff --git a/swift/common/daemon.py b/swift/common/daemon.py
index 141027c..72da20b 100644
--- a/swift/common/daemon.py
+++ b/swift/common/daemon.py
@@ -228,6 +228,8 @@ class DaemonStrategy(object):
                     self.logger.notice('Finished %s', os.getpid())
                     break
             time.sleep(0.1)
+ # force the aggregation before exiting
+ self.daemon.aggregate_recon_update()
         return 0

     def cleanup(self):

Though, I just run "swift-object-reconstructor --once", not yet run the all unit tests.