op-supervisor: cleanup, refactor, local-safe info from op-node

[protolambda]

Description

This is a cleanup PR of the op-supervisor:

• backend/db/entrydb:
  • separate the event-log-entry specifics from the basic entrydb logic
  • generalize the error types, to be reused across more DBs / functions
• backend/db/heads: remove. Cross-safety was not properly updated, and we'll need to do more work to sync L1
• backend/db/logs:
  • remove entry-index from public API.
  • entries.go to host the encoding/decoding logic that was moved out of entrydb
  • expose the timestamp information of block seals
• backend/db, ChainsDB:
  • split out queries into query.go
  • split out updates into update.go
  • RW-mutex to prevent concurrent chain set changes while querying the DBs.
  • remove db_test.go that was commented / broken.
  • and mock up queries for derived-from / finality against DB interfaces. Map entries are not initialized for now (queries will return unknown chain), but it completes the API flow to query safety data.
• backend/processors:
  • Move chain_processor.go, log_processor.go, client.go into this. For fetching and processing of unsafe blocks and their events.
• backend/safety: remove in-memory safety-index draft. It was broken in a few ways, and we need to address the design before putting something into the supervisor.
• backend/source:
  • see processors package for the things that stay.
  • dropped head_monitor.go, chain.go, head_processor.go: we are getting local-safe data from the op-node, not from geth
• backend/backend.go:
  • RW-mutex to prevent concurrent map writes during queries etc. (when adding/removing chain processors)
  • Directly maintain a set of chain-processors for the unsafe-block/event updates, no need to wrap around a chain-monitor anymore, as we're getting signals from op-node, not from geth.
• backend/mock.go: updated the mock to fit latest backend interface. It's returning zeroed values, and should be further updated for what we really need in the e2e / testing related PRs.
• frontend:
  • Add derived-from and unsafe/safe/finalized queries, to fit the op-node needs for reading data
  • Add error-return values to the update methods, to handle when the supervisor is unable to process the updates from op-node.
• types:
  • Add BlockSeal: like eth.BlockID but with timestamp. Or like eth.BlockRef but without parent-hash. Having just the seal, without parent-hash, is useful as return type in many places where we don't have the parent-hash handy.
• service.go: register the update RPC frontend.

Tests

• Updated event-logs DB tests to handle minor changes in functionality.
• Removed the commented ChainsDB tests. Work in progress, considering options on what / how to reinstate proper testing there.

Additional context

This takes a few changes from #12213

Metadata

Fix #12358

[axelKingsley]

Taking notes as I go:

• Modifications to E2E test from #12213
• Corrections to supervisor_client from #12213
• Addition of a RW lock to explicitly prevent concurrent RW
• Replacing chainMonitors with chainProcessors from #12213
• Rename db to chainsDB
• Insert Lock and defer.Unlock to new functions
• New backend functions UnsafeView, SafeView, Finalized, Update[...], Derived From.
• ChainsDB now expanded to map each L2 to a "Derived From Storage"
• ChainsDB now also has a RW lock
• entry_db now uses a type-generic T to operate on EntryTypes
• EntryTypes are uint64s which can yield to string
• entry.go now houses all the application specific entry types
• query.go and update.go now houses the ChainsDB functions

Removal of unused:

• Backend methods (old API)
• Frontend methods (old API)
• db/heads

[semgrep-app]

Semgrep found 3 sol-style-require-reason findings:

• packages/contracts-bedrock/scripts/deploy/Deploy.s.sol
  • L1002 - Triage
  • L1083 - Triage
  • L1090-1093 - Triage

require() must include a reason string

_{Ignore this finding from sol-style-require-reason.}