ansible: Try to summarize failure logs

ansible: Try to summarize failure logs

The failure logs we capture are sometimes helpful, but are often too
long, too complex, and too noisy to understand. Sentry also struggles to
associate related failures because of the presence of unique data such
as timestamps and URLs.

While I don't see a quick and generic solution to this, there are
several common failure modes that can easily be summarized. This commit
begins that work by looking for errors caused by network outages.

Signed-off-by: Zack Cerza <zack@redhat.com>