Expose liveness probe parameters for events deployment

[metrictwo]

We have large high-churn clusters with a lot of events, which occasionally causes the events pod(s) to take a while to initialize. It would be useful if the livenessProbe parameters were exposed as values so we could tune them appropriately.