get_fee_rate_statistics should aware block_ext.txs_sizes length is block_ext.txs_fees length + 1

[eval-exec]

backport #4654

What problem does this PR solve?

#4647 didn't do the right fix.

What is changed and how it works?

We should take aware:

ckb/util/types/src/core/extras.rs

Lines 13 to 30 in 44afc93

	/// TODO(doc): @quake
	#[derive(Clone, PartialEq, Default, Debug, Eq)]
	pub struct BlockExt {
	/// TODO(doc): @quake
	pub received_at: u64,
	/// TODO(doc): @quake
	pub total_difficulty: U256,
	/// TODO(doc): @quake
	pub total_uncles_count: u64,
	/// TODO(doc): @quake
	pub verified: Option<bool>,
	/// TODO(doc): @quake
	pub txs_fees: Vec<Capacity>,
	/// block txs consumed cycles
	pub cycles: Option<Vec<Cycle>>,
	/// block txs serialized sizes
	pub txs_sizes: Option<Vec<u64>>,
	}

1. BlockExt.cycles's length is equal to BlockExt.txs_fees's length.
2. BlockExt.cycles's length + 1 is equal to BlockExt.txs_sizes's length.

Because BlockTxsVerifier::verify skipped cellbase transaction:

ckb/verification/contextual/src/contextual_block_verifier.rs

Line 455 in 44afc93

.skip(1) // skip cellbase tx

But when insert BlockExt, txs_sizes have included cellbase transaction:

ckb/chain/src/verify.rs

Lines 652 to 669 in 44afc93

	Ok((cycles, cache_entries)) => {
	let txs_sizes = resolved
	.iter()
	.map(|rtx| {
	rtx.transaction.data().serialized_size_in_block() as u64
	})
	.collect();
	txn.attach_block(b)?;
	attach_block_cell(&txn, b)?;
	mmr.push(b.digest())
	.map_err(|e| InternalErrorKind::MMR.other(e))?;
	self.insert_ok_ext(
	&txn,
	&b.header().hash(),
	ext.clone(),
	Some(&cache_entries),
	Some(txs_sizes),

Related changes

• get_fee_rate_statistics should aware block_ext.txs_sizes's length is block_ext.cycles's length + 1

Check List

Tests

• Unit test
• Integration test
• Manual test (add detailed scripts or steps below)
• No code ci-runs-only: [ quick_checks,linters ]

Side effects

• None

Release note

Title Only: Include only the PR title in the release note.