Conduit: Performance Metrics Follow-Up

[shiqizng]

This is a follow-up issue to #79. After seeing the initial perf results, there's no clear explanation why the block TPS for all the runs (empty db, preloaded db) are similar. We need to collect more data to get more insights about these numbers. For the next steps, we want to

• run the same scenarios on a smaller hardware
• run a scenario for a long duration to see how the block TPS changes over time.
• Get more detailed metrics
  • Dependency: Can we integrate prometheus with DataDog?

[Will Winder]

This doesn't really describe the work needed anymore. Let's create followup issues if/when it becomes necessary with the current scheme.