Also removes the default limit of 1000, which was too low to hold more
than ~3h of L1 data. This was causing problems whenever a duration of 3h
has passed since a latest batch was posted, which then caused derivation
of a new batch to fetch all L1 data again.
This particularly impacted chains that have a usual channel duration
longer than 3h, or when chains experienced a safe head stall >3h.

Fixes #13409.

* op-supervisor: use event system like op-node, for easy synchronization

* op-supervisor,op-node: simplify local-safe derivation updates

* op-supervisor: fixes, report back problem if local-safe update fails

* op-e2e: fix fp interop action test

* op-service: bigger timeout on eventual-case in poll test

* op-supervisor: fix chain ID to index translation

* op-supervisor: fix chain index to ID translation

* feat: have AnchorStateRegistry use a single root

Updates the AnchorStateRegistry to use a single unified anchor
root by checking with the OptimismPortal for the currently
respected game type. Additionally makes the AnchorStateRegistry
MCP ready.

Users MUST deploy this contract as a new proxy and cannot upgrade
their existing proxy.

* Update snapshots post-merge

* op-deployer: Support single ASR

* Update semver

* updates after rebase

* rename method

* make isGameProper comment more obvious

* update semver

* semver bump

* simpler api for boolean functions

* comment updates

* address wildmolasses comments

* add test for not guardian

* a few more tests

---------
Co-authored-by: Matthew Slipper <me@matthewslipper.com>

* op-program: Validate that we have L2 chain configs for each rollup

* op-program: Update CLI flags to support specifying multiple L2 chains

* op-program: Allow custom l2 chain ID flag

* op-program: Implement padding steps

* op-program: Restore Head field in DerivationResult

* supervisor: Early start of RPC

* op-supervisor: Add supervisor_superRootAtTimstamp RPC

Add a new RPC to get the super root at a given timestamp.

* op-supervisor: add chainID to SuperRootResponse

* op-supervisor: fix comment

* op-supervisor: Add super root to response

* return stubbed canonical output for interop_pendingOutputV0AtTimestamp

* add ref to reorg PR

---------
Co-authored-by: Adrian Sutton <adrian@oplabs.co>

This reverts commit 66a67e02.

Seems to cause issues in CI, unclear why.
Reverting pending investigation

* op-program: Add L2 chain ID to hints when using interop

* op-program: Add L2 chain ID to OutputRoot hints when using interop

* Fix label in error message.
Co-authored-by: Inphi <mlaw2501@gmail.com>

---------
Co-authored-by: Inphi <mlaw2501@gmail.com>

* Apply Finality to blocks prior to L1 Finality Singal

* add Log when Finality is assumed from newer Finality Signal

* ensure VmBin and Server exist

* only check when not in test

* test file exists

* reuse TempDir()

* add tests for new errors

* address comments

* chore(kurtosis-devnet): split artifact management to its own package

* feat(kurtosis-devnet): add jwt tokens to output

* remove CGT code

* fixes

* fixes

* remove CGT code

* fixes

* fixes

* restore L2 tests

* make _balance a spacer, inline unnecessary internal function

* add version tests

* add version tests

* add version tests

* add version tests

* ...

* fixes

* fixes

* fix(just): properly handle multiple tags

* chore(just): rename just/ -> justfiles/

The current behavior is confusing for people who don't have `just`
installed: instead of a recognizable "command not found", they could
see a "Permission denied" (if . is in the PATH I suppose).

* op-program: Introduce L2Sources to combine info for multiple L2s.

* op-program: Update the host config to support multiple L2s

The CLI flags still only support a single L2.

* op-program: Update interop bootstrap to load multiple chain configs

op-program: Introduce RetryingL2Sources and use it to make prefetcher somewhat multi-L2 capable (#13718)

* op-program: Introduce L2Sources to combine info for multiple L2s.

* op-program: Plumb L2Sources in

* op-e2e: Use actual chain ID, not hte test place holder.

* op-program: Use correct chain ID for rexec and prevent duplicate RPC urls for same chain

* op-program: Fix test to use chain ID that is actually configured in the test

* ctb: Make implementations deployments deterministic and idempotent

* code review updates

* cr updates

* imports

* make superchain deployments deterministic

* forgot to add test

* code review updates

* broadcast in dpeloysuperchain

* add forgotten broadcast

* op-program: Pass required block hash to OutputByRoot rather than to L2Source constructor

* Move check that output root matches requested hash to prefetcher instead of source.

* chore: bump kontrol version, and make output checks more robust

* chore: remove outdated block

* fix(kurtosis-devnet): ensure tests pass without kurtosis running

* fix(kurtosis-devnet): forbid tests from using local kurtosis context

make the unused imports script check support multiline formatted imports, remove unused imports (#13695)

Previously it always picked the first block which was the worst case pre-holocene.
Now it biases towards the first and last blocks as they hit more corner cases but
can pick a random block in the middle of the batch as well.

* op-program: Support agreed prestate hint in host

* op-program: Fix transition state serialization in test

* op-program: Fix prefetcher_test.go

* op-program: Prevent specifying both --l2.outputroot and --l2.agreed-prestate flags

Check config is consistent

* op-program: Return safe or local safe head from derivation

Avoids relying on the safe label which won't be updated on interop chains.

* op-program: Only stop derivation if local safe reaches the target block.

* op-program: Fix safe head trace extension

* op-program: Remove unused field.