Identify when the inspected server is restarted, and terminate

Context: 2nd scenario depicted in https://github.com/github/gh-ost/issues/784

In the 2nd scenario, what happened was that the inspected replica was rebooted. `gh-ost` kept hanging onto the end of the binary log where the replica was rebooted, and never rotated onto the next binlog. It also apparently retried enough times to see the replica healthy again.

`gh-ost` should identify when a server is restarted. At this time I'm not expecting it to recover and resume work -- instead I prefer that it terminates with error.