Improve blockfrost chain following approach

[noonio]

Description

The current implementation Hydra.Blockfrost.ChainObserver is using the following algorithm to "follow" the chain via Blockfrost:

1. GET /block/{hash}
2. GET /blocks/{hash}/txs
3. for each -> GET /txs/{hash}/cbor
4. process & take maybe nextHash from latest known block (1)
5. GET /block/{hash}

As part of this general process:

Issues we want to fix

• Our queries are inefficient as it requires lots of API calls.
• We do not handle rollbacks at present
  • Test to see what our present code does under rollbacks (may require a bit of hacking/mocking to get this working)

Suggested solution

• Design the ideal "following" logic for Hydra node
• Propose and implement changes to the Blockfrost API that are required
• Implement the logic in Hydra node and check that it functions the same (or better!)

Alternative

• Leave our logic as-is; it's inefficient but works.

Additional context

No response

Todo

• Include issues we opened against the blockfrost API ( cc @ffakenz )

[ch1bo]

@mmahut Let's discuss possible ways to get this done and this would also be a good "entry point" into the hydra code base and is mostly cardano specific!