Preserve published layers

Storyblok stories can have two meaningful content layers:

• draft/current content returned by the Management API
• published production content from Story Versions API

When a story is published and has unpublished changes, those layers are intentionally different.

The problem

A single-layer migration has to choose a bad tradeoff:

• publish the latest draft and expose editor work early
• save the migrated draft and leave production JSON unmigrated

--publicationMode preserve-layers avoids that tradeoff for in-place space migrations.

Command

sb-mig migrate content --all \
  --from 12345 \
  --to 12345 \
  --migration migration-name \
  --publicationMode preserve-layers \
  --publicationLanguages all \
  --dry-run

preserve-layers currently requires:

• --migrate-from space
• the same Storyblok space for --from and --to

Dirty published write order

For stories where published === true and unpublished_changes === true, sb-mig:

1. reads draft/current content from Management API
2. reads the latest status === "published" version content
3. migrates both layers independently in memory
4. writes migrated published-layer content
5. publishes that migrated published layer
6. restores migrated draft/current content with publish:false

The final state keeps the production and draft split:

published JSON: migrated production content
draft JSON:     migrated editor draft content
story state:    published=true and unpublished_changes=true

Safety checks

Before the dual-layer write, sb-mig re-fetches the story and compares current_version_id or updated_at. If the story changed after migration input was read, it refuses to overwrite new editor work.