bug: taker fee ingest is coupled to pools

[p0mvn]

An issue was observed where a taker fee for WBTC > allBTC is set to 0 (confirmed via CLI query) but the swap widget still shows it as 0.1%.

This is the investigation summary:

As of my knowledge, there is no cache for taker fees at any level of abstractions. Did a quick skim to validate that. The FE simply fetches the result from SQS quote after converting swap fee into the fiat value.
What seems to be the issue here is the way we propagate taker fee updates into SQS.
The way the ingestion works is that it attempts to retrieve for all combinations of denoms within pools that were modified in a block.
If there was no state update over the alloyed BTC pool for 20 minutes, the taker fee update would not get picked up.
This logic stems from the original implementation that pushed all data every block so this wasn't a problem.
Longer term, we should decouple the taker fee propagation from pool propagation. However, this feels like a low priority issue.

DoD

• Decouple taker fee ingest from pool ingest
• Unit test
• Integration test