count sync errors

[jacobstr]

What this PR does / why we need it

It adds metrics to count failures in the sync reconciliation loop of the certificates controller. I'd like visibility into e.g. failure renewing certificates. I see that there's a similar MR to count sync operations across controllers (cert-manager#1692). That one looks like it casts a wider net but doesn't capture whether an error occurred and why it occurred.

Reviewer Notes
The implementation feels a bit cumbersome, wrapping the c.issue calls up felt better than if err / else blocks throughout the relevant lines.

Alternate Approach: Convert Events to Metrics

The controller does emit events. If there's a general cluster component that converts events into metrics, one could use that to generate metrics from the events the controller already emits. There is kube-state-metrics, but it doesn't expose event data - possibly because events are a poor match for "state".