Benchmark `validate_data_column_sidecar_for_gossip ` for gossip verification performance

[jimmygchen]

We should benchmark validate_data_column_sidecar_for_gossip :

lighthouse/beacon_node/beacon_chain/src/data_column_verification.rs

Line 458 in e6ef644

pub fn validate_data_column_sidecar_for_gossip<T: BeaconChainTypes, O: ObservationStrategy>(

• Gossip verification is a critical path for block and blob propagation.

• The introduction of data columns significantly increased verification volume (e.g., bursts of 128 sidecars) (called per sidecar)

• Gossip verification has been observed taking longer than expected on PeerDAS devnets:

  • Data column gossip verification taking more than 4s when blob count is above 20 #7203

[gitToki]

Hi @jimmygchen, we would be interested to work on that with @leobago

[jimmygchen]

That'd be great, thanks @gitToki ☺

[jimmygchen]

I did a bit of testing and got a breakdown here:
https://hackmd.io/5qRLqpJaSFy7DOHSuyJWxw?view#gossip_data_column

The finding suggested:

1. Gossip verification time metric on Grafana was a bit inflated because of how the buckets and the grafana query are set up.
2. Most of the time is spent on KZG verification, which we can optimise by batching them.

@gitToki I think we may not need to create benchmark code for this, because we could just get real data with tracing spans, which I plan to work on in #7709.