fix(consensus): reconcile existing safe block before build fallback

[niran]

Summary

• check for an already-imported derived safe block before falling back to FCU-with-attributes build
• when the block exists and matches the derived attributes, apply it through the follow-safe path so unsafe/local-safe/safe advance together
• treat follow-safe FCUs that return without advancing state as temporary errors so the consolidate task is retried instead of silently popped
• add regression coverage for the case where an earlier unsafe FCU returned SYNCING, EL later backfilled the block, and safe derivation resumes while consensus unsafe is still pinned

Testing

• cargo +nightly fmt
• git diff --check
• cargo test -p base-consensus-engine

Change Management

type=routine
risk=low
impact=sev5

[cb-heimdall]

🟡 Heimdall Review Status

Requirement	Status	More Info
Reviews	🟡 0/1	Denominator calculation Show calculation 1 if user is bot 0 1 if user is external 0 2 if repo is sensitive 0 From .codeflow.yml 1 Additional review requirements Show calculation Max 0 0 From CODEOWNERS 0 Global minimum 0 Max 1 1 1 if commit is unverified 0 Sum 1

[github-actions]

nit: The error from l2_block_by_label is discarded here. While this follows the pre-existing pattern at line 262, it makes diagnosing RPC failures harder. Consider logging the error:

Suggested change

	Err(_) => {
	warn!(target: "engine", block_num, "Failed to fetch derived L2 block before build fallback");
	return Err(ConsolidateTaskError::FailedToFetchUnsafeL2Block);
	Err(e) => {
	warn!(target: "engine", block_num, error = ?e, "Failed to fetch derived L2 block before build fallback");
	return Err(ConsolidateTaskError::FailedToFetchUnsafeL2Block);
	}

[github-actions]

✅ base-std fork tests: all 616 passed

base/base is fully in sync with the base-std spec.

Dependency	Ref	Commit
base-std	main	4658f1b7
base-anvil	0092692587d8d064dd2c6923ce26a682c58f3694	00926925

[github-actions]

Review Summary

Overall this is a well-structured fix for a real edge case where an earlier unsafe FCU returns SYNCING, reth backfills the block, but consensus never advances its unsafe head — causing safe derivation to fall through to an unnecessary build_and_seal.

What the PR does right

• The new reconcile_existing_derived_block method correctly checks whether the EL already has the derived block before falling back to FCU-with-attributes build, avoiding stale builds.
• The ForkchoiceUpdateDidNotAdvance post-condition check in reconcile_to_safe_head correctly catches the case where SynchronizeTask succeeds but the state didn't actually advance (e.g., EL returned SYNCING and safe_only_sync_update filtered the update). Marking this as Temporary severity ensures the task is retried rather than silently dropped.
• The execute() refactor from a flat if/else to a match correctly preserves the original semantics while inserting the new reconciliation step only in the Attributes path where safe_head >= unsafe_head.
• Tests cover both non-span and last-in-span variants and verify that no FCU-with-attributes (stale build) is issued during reconciliation.

Minor finding

• Discarded error context (task.rs:207): The Err(_) in reconcile_existing_derived_block discards the actual RPC error, making production debugging harder. This follows the pre-existing pattern but would benefit from logging the error value. (Left inline comment with suggestion.)