With interop, the op-supervisor drives the node to sync the right things.

Heading into the upgrade, we need pre-interop derivation to run, and then transition into managed derivation.

We should make the op-supervisor drive, but not put any of the pre-interop data into the DB

Changes:

• sync-node in supervisor should not be blocked by the anchor-point; the anchor point will not exist pre-interop
• data that is received from the op-node should be filtered; pre-interop data should not be sent off as event for DB updating
• the supervisor should retrieve the anchor-point when it's time for interop, and seamlessly transition to start indexing the chain.

Additionally, we need:

• apply additional boundary checks on the DB to enforce the starting point.
• handle anchor-point in expiry case, when the start of the DB is too old. Do we want to sync from non-0, when past expiry? Something like starting at max(upgrade_timestamp, current_timestamp - history_size). A: always sync from anchor, even if expired log data. But never before interop activation block.