Release dormant Ghost content identity primitives

[aileen]

Parent

• PRD #213 — Deliver the maintained HTML-to-Algolia extraction pipeline
• User stories covered: 34–36, 48–49, 51–52

What to build

Add and release the library primitives required for pages without activating any CLI or webhook behavior. The fragmenter produces explicit type-tagged Ghost content identity through its deep API, and the indexer can browse, replace, delete, and reconcile package-managed records by that identity while old slug APIs and defaults remain available.

Acceptance criteria

• The deep fragmenter API accepts explicit Ghost content type and stable Ghost content ID from a trusted adapter and repeats protected contentId and contentType in every derived record.
• Ghost content type is exactly post or page; it is never inferred from an incomplete reduced object.
• Physical fragment object IDs remain deterministic Algolia upsert identities while content ID becomes the logical grouping/replacement/deletion identity.
• The indexer exposes public content-identity browse, replace, delete, and reconciliation operations that preserve unrelated index records.
• Existing slug operations, posts-only defaults, and current adapter behavior remain usable and unchanged.
• Public tests cover same-slug/different-type records, multi-fragment identity, replacement after shrink, deletion, pagination/classification inputs, escaping, error mapping, and unrelated-record preservation.
• The new identity fields remain subject to the complete 9,999-byte record ceiling.
• Fragmenter and indexer minors are selected together in a scoped dry run; dependent application patches are documented as dependency-only.
• Published artifacts are clean-install verified and release notes clearly state that the APIs are dormant: installing them does not enable pages or change source selection.

Blocked by

• #223 — Release richer searchable rendered meaning