[P1][DATA-01] Entities, observations and evidence

Programme: #1
Epic: #17

## Objective
Create the semantic substrate that lets services act as data/capability providers without collapsing provider-specific truth into one lossy universal schema.

## Scope
- Small common ontology: Person, Organization, Product, Offer, Article, Claim, Post, Thread, Message, Event, File, Task, ActionReceipt plus provider extensions.
- Stable internal IDs plus provider object/account IDs.
- `Observation` records with source URI/object, source revision, observed/source timestamps, extractor version and content hash.
- `Evidence` spans linking claims/fields to bounded source material.
- Access, sensitivity and freshness labels.
- Separate user overlays and derived/inferred claims from source observations.

## Data rules
- Product != offer != order; person != account; article != claim.
- Provider originals and identifiers are preserved.
- Inference cannot silently overwrite authoritative source fields.

## Acceptance criteria
- [ ] Source observations, user overlays and inferred claims are distinct and independently queryable.
- [ ] Each critical source-backed field can expose provider object ID, observation time and freshness.
- [ ] Provider-specific fields can coexist without polluting the common schema.
- [ ] Account/workspace scope is part of entity/observation identity and cache keys.
- [ ] Evidence spans survive normal migrations and are invalidated when the referenced source revision changes incompatibly.

## Dependencies
- CORE-01
- CORE-02
- SEC-02

**First phase:** P1
**Maturity target:** P2
**Owner:** workspace-ui-data

## Task issues
- [ ] #344 `DATA-01.T01` — Define the common ontology and extension rules
- [ ] #345 `DATA-01.T02` — Implement scoped identity and entity resolution
- [ ] #346 `DATA-01.T03` — Build immutable observation ingestion
- [ ] #347 `DATA-01.T04` — Implement evidence spans and claim associations
- [ ] #348 `DATA-01.T05` — Separate overlays and inferred projections
- [ ] #349 `DATA-01.T06` — Integrate access, freshness and retention metadata
- [ ] #350 `DATA-01.T07` — Publish normalization conformance fixtures

Start by reading the listed dependencies (#1, #17, CORE-01, CORE-02 and SEC-02), then use task issues #344–#350 to break down the ontology, identity, observations, evidence, overlays, metadata and fixtures work. Done means meeting all five acceptance criteria while preserving provider-specific fields, scoped identity and evidence validity.

Written by the indexing model from the issue text.