early crash/non-response detection

Problem

At the moment, DNS Shotgun does not try detect/act if the other side is not responding at all and the test always runs to the end of input file.

Scenarios

It would be beneficial if:

• Server under test crashed
• Server under test did not even start (misconfiguration)
• Network does not pass packets back and forth (misconfiguration)

Proposal

Detect large blocks of consecutive timeouts and gracefully terminate test if that happens. This is implemented in respdiff, so I think the same logic could be re-purposed for DNS Shotgun.