In DPTP-2921 we've established ongoing issues with disruption triggering on DNS problems in the actual cluster running the tests, not the cluster being tested.

These disruption intervals have a very defined error message that we believe can be safely assumed to be not real disruption: dial tcp: lookup HOST: i/o timeout.

Update origin disruption code to detect this message, lower interval severity to warning which should remove it from consideration in the tests and the disruption json files we output, and still graph it on the intervals chart.