Related: #146

The 10k user scale test should provide data for scaling formulae and performance enhancements for months after the test. We can use the in-product Prometheus metrics to collect performance data while the test(s) run.

This means we need a long-lived Prometheus instance for the duration of our scaletesting development and execution. This will scrape snapshots from the control plane during tests and stay alive in between them.

Instead of automated bucketing or any automation around scaling, we'll keep it simple and generously provision the disk on this instance.

This fortunately simulates real-world data collection load that an administrator may conduct in production.