The agentless metric will continue to poll on the interval that is set for it, regardless of the current status of the rule. Unlike OpenView or Nagios however, there is not functionality within the monitoring tool to change the frequency of tests based on the state of a rule. As for email alerting, this can be handled under the actions/severities related to the policy.
As an example, in the agentless ping metric, you have the polling set to 15 seconds. You have a rule that evaluates that metric, and on an unavailble result from the metric, you set the severity in the rule to critical.
At the policy level, under the actions tab, you associate an email task both critical and normal severity. During a state change between normal to critical, you will then receive an email, as well as a state change from critical to normal.
You cannot, as I mentioned, change the metric's gathering frequency based on the rule's current state, so having it poll more frequently when it is in a bad state is currently impossible.