op-node/rollup: Promote all attributes to safe post-Holocene

[sebastianst]

Description

Holocene removes backwards-invalidation by span batches. So all derived attributes can immediately be promoted to safe. This is the only change in logic of this PR.

The PR also renames fields IsLastInSpan to Concluding in many structs because that's semantically clearer. That a last batch in a span batch is pre-Holocene the indicator for a safe block promotion is an implementation detail. It's better to name the effect rather than the reason. Concluding was chosen, since these attributes conclude the pending phase.

Tests

Changed an assertion in a test from using pending safe to safe to test this change. It failed before.

Additional context

Holocene skips overlapping span batch checks, which in a constructed malicious sequencer setting can lead to a local chain split of nodes that force-restarted mid-way through block derivation. However, the same situation can already be applied pre-Holocene with singular batches, because they are also fast-dropped and overlapping singular batches are not tested to match an existing safe chain. So we may actually want to change the mechanics of safe chain promotion from pending safe to only promote after a full L1 block has been derived. This is reserved for a follow, more details can be found here: #12695

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 63.85%. Comparing base (5e8dbc0) to head (3313459).
Report is 92 commits behind head on seb/holocene-invalid-payloads.

Additional details and impacted files

@@                        Coverage Diff                        @@
##           seb/holocene-invalid-payloads   #12724      +/-   ##
=================================================================
- Coverage                          64.86%   63.85%   -1.02%     
=================================================================
  Files                                 54       55       +1     
  Lines                               4460     4606     +146     
=================================================================
+ Hits                                2893     2941      +48     
- Misses                              1391     1483      +92     
- Partials                             176      182       +6     

Flag	Coverage Δ
cannon-go-tests	63.85% <ø> (-1.02%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

see 1 file with indirect coverage changes

[sebastianst]

Marked the actual changes. Everything else is just renaming IsLastInSpan to Safe.

[protolambda]

The PR also renames fields IsLastInSpan to Safe in many structs because that's semantically clearer. That a last batch in a span batch is pre-Holocene the indicator for a safe block promotion is an implementation detail. It's better to name the effect rather than the reason. And in some deriver events, it was already called Safe.

While I generally agree with the effect vs reason naming, unfortunately it does not work so well here:

• I find "safe" to be confusing with the "pending safe" concept.
• "safe" (as RPC label) is equal to cross-safe in interop, whereas deriving it for the chain without cross-L2 checks is only considered "local safe" in interop, and thus shouldn't be confused with "safe"

And so since the functionality is actively changing, the effect is not the same across forks, and thus I much prefer the naming based on the reason.

Alternatively, we can name it "stableDerived" or maybe just "stable", or something that makes it clear it's not a pending attribute (nor a sequencer unsafe attribute). (Named after the original 'stable batch derivation' feature name)