core: remove withdrawal length check for state processor

[islishude]

I thought the withdrawal length check should be placed in BlockValidator.ValidateBody

go-ethereum/core/block_validator.go

Lines 73 to 84 in 32a1e06

	if header.WithdrawalsHash != nil {
	// Withdrawals list must be present in body after Shanghai.
	if block.Withdrawals() == nil {
	return errors.New("missing withdrawals in block body")
	}
	if hash := types.DeriveSha(block.Withdrawals(), trie.NewStackTrie(nil)); hash != *header.WithdrawalsHash {
	return fmt.Errorf("withdrawals root hash mismatch (header value %x, calculated %x)", *header.WithdrawalsHash, hash)
	}
	} else if block.Withdrawals() != nil {
	// Withdrawals are not allowed prior to Shanghai fork
	return errors.New("withdrawals present in block body")
	}

the BlockValidator has the chain config field, and should be feasible to move the check to the ValidateBody.

but there is only a withdrawal root hash check.

and the verifyHeader func in the consensus package ensures a block that before shanghai fork doesn't have a withdrawal hash.

go-ethereum/consensus/beacon/consensus.go

Lines 267 to 273 in 32a1e06

	shanghai := chain.Config().IsShanghai(header.Number, header.Time)
	if shanghai && header.WithdrawalsHash == nil {
	return errors.New("missing withdrawalsHash")
	}
	if !shanghai && header.WithdrawalsHash != nil {
	return fmt.Errorf("invalid withdrawalsHash: have %x, expected nil", header.WithdrawalsHash)
	}

so I think the check in the processor is not required and can be removed.

[rjl493456442]

Personally I think the change is good, but i would prefer to discuss with team tomorrow as it's a very core part.

[holiman]

I can think of no reason to have this check here, that would not be covered already by the other checks. LGTM