FreeNAS Code

Insert the snapshot we found to want_list or we miss it forever.

- Increase scope for mount lock to avoid race condition when two
replicator kicks in at the same time;
- Log mismatch after replication, which is usually a fail condition;
- Break for a file system after failed replication rather than trying
newer ones.

- Add more logs;
- Fix pipe operation so that all outputs goes to the temporary log file;
- Don't catch exceptions other than OSError when killing - exit() is
implemented by catching SystemExit.

Add a 2 seconds' delay before we start replication.

This is not ideal, a better solution is being worked on.

Consistently use timedelta instead of just replacing components.

This fixes #492.

Catch a case where snapshot is gone remotely while we have a "LATEST"
snapshot.

The beginning of the end.

add field to limit replicationm speed using throttle pipe;
add userepl field in Replication to migration as well, seems forgotten

Checkpoint. Don't use this.

Add an option to use the '-R' option of zfs send.

While I'm there fix the issue that if the remote side deleted all
snapshots, the replication doesn't start over.