Update indexer rev to sleep time between request and Hasura metadata loading

[musitdev]

Summary

• RFCs: Link to RFC, Link to RFC, or $\emptyset$.
• Categories: any of protocol-units, networks, scripts, util, cicd, or misc.
• Associated PRs: remove panic in write_set_change grpc convertion aptos-core#66

Define the env var SLEEP_TIME_BETWENN_REQUEST_MS to define the sleep time between indexer gRpc request.
By default, it's set to 10 ms.
If SLEEP_TIME_BETWENN_REQUEST_MS is et to zero, no sleep is applied between request.

Add a command to load the Hasura metadata.
The meta data file is localed in the folder networks/suzuka/indexer and the metadata file is : hasura_metadata.json

I've updated the Aptos one.
The file must contain the IP address of the PostgreSQL db so that Hasura can load it. To update dynamically the file with the DB IP and load the metadata, I've created the load_metadata binary that can be run this way:
POSTGRES_DB_HOST=<POSTGRES_HOST_IP> INDEXER_API_URL=<hasura console url> cargo run -p suzuka-indexer-service --bin load_metadata

• POSTGRES_DB_HOST: set to the IP of the PostgreSQL DB, default postgres
• INDEXER_API_URL: Url of the Hasura console, default: http://127.0.0.1:8085

Changelog

• Add SLEEP_TIME_BETWENN_REQUEST_MS to slow indexer request
• Add the load_metadata binary to load Hasura metadata.

Testing

To test the whole chain locally:
CELESTIA_LOG_LEVEL=FATAL nix develop --extra-experimental-features nix-command --extra-experimental-features flakes --command bash -c "just suzuka-full-node native build.celestia-local.indexer.hasura.indexer-test --keep-tui"

Outstanding issues

[l-monninger]

@musitdev Can you make sure this passes before I review?

[musitdev]

@l-monninger the test pass now, it made a mistake and the test never end. So I split the 2 in a separate one.
I've updated the PR description for the second time. The description was reverted to the initial one!
This PR contains 2 evolutions: the time between indexer request and the Hasura metadata loading and indexer test update.
You can review the PR.

[musitdev]

@l-monninger I made an update and restart the test. I tell you when it's done.

[musitdev]

@l-monninger the test has passed.

[l-monninger]

Working for me. But, did you use a dynamic environment to see if this improves those early crashes? Can you link that?

[l-monninger]

Would be better to move this into a crate at some point.

[musitdev]

You prefer to implement it in Rust?

[musitdev]

Yes, I test it to synchronize the current suzuka db. The diff is that the suzuka node wasn't processing Tx. It synchronized more than 130M Tx and never crash. I've tested with 10ms and 1ms delay.