Logic to prevent double counting of DefiLlama TVL assets

[ccerv1]

What is it?

OP Labs has implemented as logic to handle this, which they are willing to share. Ideally, we ingest (or replicate) their model.

For more context, see this notebook. It shows how "borrowed" and "staked" versions of assets can be double counted.

We should create an intermediate model (eg, int_superchain_defillama_tvl_filtered) that cleans this up and have the OP labs team review it / compare with their own TVL metrics.

[linear]

OSO-379 Logic to prevent double counting of DefiLlama TVL assets