fix: delete retry requests if confirmed or sent

[jhkimqd]

This PR addresses a rare condition where some transactions' status changes (possibly due to network issues) and causes a memory leak in the pool-manager which was causing the system to crash. Although a restart of the pool-manager fixed the issue, after applying the changes in this PR, the same issue could no longer be reproduced.

Testing

The entire test had been conducted using Kurtosis CDK and loads were sent using polycli loadtest, using hermeznetwork/zkevm-pool-manager:v0.1.0

When sending moderate load (10,000 EOA transfers) to the CDK Erigon RPC node, it directs the transactions to the pool manager. The loadtest also involved 429 codes returned by the Erigon RPC, and had some transactions which had unexpected status values in the pool_manager_db

These were not mined - until a replacement transaction has been sent with higher gas prices.

This seemed to cause the pool-manager to detect these transactions and attempt to delete and enqueue them again - which seemed to be the source of the leak.

With this PR, the issue could no longer be reproduced under identical conditions.

[sonarcloud]

Quality Gate failed

Failed conditions
B Maintainability Rating on New Code (required ≥ A)

See analysis details on SonarCloud

Catch issues before they fail your Quality Gate with our IDE extension SonarLint

[vcastellm]

LGTM