Sync composability indeed requires the L1 for finality, and that causes issues between Zones if not carefully designed.

A multi-zone transactions needs to be settled on-chain, and must consume the UTXOs of the involved Zones.

If there is a reorg, and that "merging" UTXO disappears, then the history of those Zones is rewinded back to that point.

In this scenario, as long as it is happening synchronously for the involved Zones, new executors will need to rebuild. This is only expected in the case of an attack, such as long-range attack, as forks should be resolved immediately otherwise.

The problem here becomes of data availability (for the executors to rebuild). This is why we will benefit from multiple levels of availability and we should take into account L1 "weak subjectivity" period for the DA redundancy.

An Ultra-optimized path

Sync composability with atomic inclusion implies consuming the UTXO of 2+ zones simultaneously.