IntegrationTest: Starting and stopping many nodes, seeing them get downed etc

A "stress integration test" with harshly killing -9 nodes and seeing that the rest of the cluster handles it nicely.

The issue does not name a file or existing test entry point. Start by locating the integration tests covering cluster node startup and shutdown, then determine how nodes can be repeatedly started, stopped, and harshly killed with -9. Done means the test exercises these failures and verifies that the remaining cluster continues to handle them correctly.

Written by the indexing model from the issue text.