fix(lean): bound block slot before the state transition in on_block

[zemse]

What was wrong?

Based on leanEthereum/leanSpec#1182

Proposer for a block is slot % num_validators. So a malicious proposer can sign a valid lean block at high slot number where block.slot % num_validators == validator_index.

The on_block has no upper bound on the block.slot and when such a malicious block is imported, the following while loop runs for a huge number of iterations.

ream/crates/common/consensus/lean/src/state.rs

Line 118 in c30a9c9

while self.slot < target_slot {

While it rejects later in process_block_header however honest validator has already burned CPU on high number of iterations due to the malicious block.

How was it fixed?

Two guards are added in the on_block so that such a malicious block is rejected before it gets into state_transition call.

• Parent-gap cap: reject block.slot − parent.slot > MAX_HISTORICAL_BLOCK_HASHES i.e. 2^18.
• Clock horizon: reject block.slot > current_slot + 1, keeping far-future blocks out of the store.

To-Do

• I have read CONTRIBUTING.md.
• This PR title follows conventional commits.

[Kayden-ML]

@zemse can you rebase your pr

[zemse]

Just rebased over latest master

[Kayden-ML]

Looks good 👍