Reorder the db WriteBlock codes to reduce the harm of partial-write

[DarianShawn]

Description

The PR fixes a fatal problem when block is partial written to the db.

When a node is Writing block, the state was written after WriteHeader, once the node restart just after WriteHeader, the node would definitely failed to sync from other nodes any more.

The only way to recover by now, is to remove all blockchain data, and restart syncing from the Genesis block.

Because the node take if for granted that it has the the current header's state saved, though it is not. When syncing blocks, it would failed due to state not found.

The best way would be using batch written implementation like go-ethereum, which provides some transaction-like mechanism. It should be another big PR though.

Changes include

• Bugfix (non-breaking change that solves an issue)
• Hotfix (change that solves an urgent issue, and requires immediate attention)
• New feature (non-breaking change that adds functionality)
• Breaking change (change that is not backwards-compatible and/or changes current functionality)

Checklist

• I have assigned this PR to myself
• I have added at least 1 reviewer
• I have added the relevant labels
• I have updated the official documentation
• I have added sufficient documentation in code

Testing

• I have tested this code with the official test suite
• I have tested this code manually

Manual tests

1. Set up a 4-validators network.
2. Stop a normal validator, wait a minute or so.
3. Replace with a reproduction node
4. Replace the normal node back.
5. We should see logs like jury-bulksync-fail-202205024.log

[abrahamcruise321]

LGTM