Improve AvalancheGo Re-Execution Benchmarking

[aaronbuchwald]

This ticket expands on all of the potential future work outlined in #4019.

Improved tooling for data collection + benchmarking:

• Fetch block range from the network and archive to S3 #4105
• Add generic VM re-execution benchmark support (include P-Chain/X-Chain) #4108
• Add optional feature to archive end-state of re-execution to S3 ie. execute range [10m, 20m] and archive state as of 20m #4167
• Update Grafana Link to use Accurate Start/End Times based on Run Duration #4390
• Add worst case mgas/s over 15s (or default to use lowest data point ie. scrape interval) to top-level metrics #4391

Improve benchmark/testing:

• Identify re-execution micro-benchmark(s) that can run on every PR #4109
• Support Re-Executing Chain History in Parallel using Sub-Ranges #4111
• Configure CI definitions internally to the task definition

Further Test Cases

• Hashdb full node
• Hashdb archive mode
• Firewood
• Add Re-Execution Benchmarks for Pathdb Config #4233

Documentation:

• Provide a "How-To Guide" for custom VM testing using the same approach

[aaronbuchwald]

Some feedback from @qusuyan

• Want the ability to pass in config via CLI (currently hardcoded and this is outlined in Add generic VM re-execution benchmark support (include P-Chain/X-Chain) #4108)
• Want the ability to toggle metrics-enabled flag

[aaronbuchwald]

Some feedback from @qusuyan

• Want the ability to pass in config via CLI (currently hardcoded and this is outlined in Add generic VM re-execution benchmark support (include P-Chain/X-Chain) #4108)
• Want the ability to toggle metrics-enabled flag

These are completed by:

• Add support for passing config and predefined configs to VM re-execution tests #4180
• feat: add parameters to disable metrics #4214