Build Grafana dashboard to compare Puma and Unicorn on Ops environments

We should have a single dashboard that could be used to compare Puma and Unicorn after switch in their behavior.

We want to understand the impact on:

1. CPU usage pattern,
2. Per-CPU utilisation,
3. System-memory usage,
4. Per-worker memory usage pattern,
5. Rate of workers (Unicorn/Puma) restarts,
6. GC cycles,
7. GC duration,
8. Memory utilisation (amount of empty slots vs allocated memory, as we want to recycle as much as possible),
9. Connections queued and queue latency,
10. Active connections.