cbbackupmgr should cleanly upload the logs to S3

At the moment the logs are uploaded to S3 before the '(Cmd) success' log line; this will make parsing the logs with tools like "timeline" slightly more difficult. We should upload all of the archive metadata (excluding the logs) then finally we should upload the logs. This allows us to have archive logs which will match those created when doing a backup locally.

This can be done with a lot less hassle with the new 'exitIfError' function; passing in an object store client will allow use attempt to upload the logs before we close the archive and remove the lockfile.