Patch moonbase block

[RomarQ]

What does it do?

This PR fixes a problem that was caused by a change in frontier, which moved Pending::<T>::kill() from on_initialize to on_finalize: polkadot-evm/frontier#638

The bug was included with runtime 1603 in block 2285347, which caused block 2285348 to include transactions from the previous block.

What important points reviewers should know?

This is a client specific fix, which just checks if the block being queried is the one affected and filters out the invalid transactions.

Is there something left for follow-up PRs?

• Add smoke test

[crystalin]

@RomarQ I think we decided not to fix it as it required a permanent patch and wasn't worth it.
Do you think it is worth? What was the reason you need it?

[RomarQ]

@RomarQ I think we decided not to fix it as it required a permanent patch and wasn't worth it. Do you think it is worth? What was the reason you need it?

A user opened an issue about it: #3026, it seems to be causing issues in their application.

[crystalin]

Ok, Can you evaluate the cost of maintaining this patch long term vs just maintaining that knowledge (and replying to possible rising issues in the future) ?
(It might already be included in our inconsistant state doc: https://docs.moonbeam.network/builders/build/historical-updates/)

[RomarQ]

Ok, Can you evaluate the cost of maintaining this patch long term vs just maintaining that knowledge (and replying to possible rising issues in the future) ? (It might already be included in our inconsistant state doc: https://docs.moonbeam.network/builders/build/historical-updates/)

Yes, it is documented there already: https://docs.moonbeam.network/builders/build/historical-updates/#ethereum-transactions-duplicated-in-storage

Since the problem cannot be solved entirely without an hard-fork, I think it will be better to just keep the issue documented and expect people to handle these inconsistencies on their side.