docs: reorganize tree, restore architecture + research from archive, add Phase 0 reports

Active docs/ was stub scaffolding while the real docs sat in docs/_archive/.
Restore and correct them, and record the Phase 0 work.

- docs/architecture/current.md: verified system map (from archived SYSTEM_OVERVIEW,
  9 corrections against code).
- docs/research/competitors.md: sourced competitor analysis (from archived
  PRODUCT_RESEARCH, feature matrix corrected).
- docs/decisions/ADR-002-job-application-model.md: the Job/JobApplication split.
- docs/application-discovery-report.md, docs/implementation-roadmap.md,
  docs/phase-0-foundation-report.md, docs/career-workspace-branch-assessment.md.
- Remove 10 zero-byte placeholder files that advertised content that never existed.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

docs/_archive/ARCHITECTURE_EVOLUTION.md

@@ -0,0 +1,53 @@
+# Jobbjakt Architecture Evolution
+
+## Guiding constraints
+
+- Preserve the existing single-master-profile model.
+- Keep job applications and their workflow at the centre of the domain.
+- Make AI providers replaceable and optional.
+- Add SaaS foundations without prematurely building a billing platform or team product.
+- Keep feature slices small while reducing the current controller concentration.
+
+## Target boundaries
+
+```text
+Web UI
+  -> Application APIs
+      -> Application workflow domain
+      -> Career profile and document domain
+      -> Discovery provider adapters
+      -> Connected-account adapters
+      -> AI provider boundary
+      -> Entitlements and usage policy
+```
+
+### Career profile and documents
+
+`ApplicationUser.ProfileCvStructureJson` remains the canonical master profile during the transition. A later migration may normalize high-change sections (projects and portfolio items) into user-owned tables, but generated CVs and application-specific drafts must remain immutable copies with a profile-version reference.
+
+### Discovery
+
+Create an `IJobDiscoveryProvider` contract with normalized query and result types. Providers own authentication, source-specific filters, terms-aware attribution, and mapping. The UI and import flow must not know provider payload shapes. Cache public results briefly, do not persist an external job until the user imports it.
+
+### AI
+
+Extract AI prompts and model transport from controllers behind task-focused interfaces: CV extraction, CV tailoring, draft generation, interview preparation, and summarization. Each request records provider, model, input profile version, job context, and user review state. Local and online providers implement the same boundary.
+
+### Entitlements
+
+Introduce account-level plan, entitlement, and usage-ledger records before payment integration. Product code checks named capabilities (for example `ai.tailor_cv`, `cv.premium_theme`, `storage.attachments`) rather than payment-provider state. This supports trials, manual access, subscriptions, and future organisation plans.
+
+## Incremental engineering work
+
+1. Add targeted feature services beside the existing controllers for every touched feature; do not attempt a big-bang controller rewrite.
+2. Replace local-only preferences that affect behaviour with user-owned persisted settings.
+3. Replace the in-memory CV queue with a database-backed work item before horizontal hosting.
+4. Use PostgreSQL for hosted multi-tenant deployment; retain SQLite for self-hosted/single-user deployments.
+5. Add provider health, job processing, AI usage, and entitlement audit telemetry before paid hosting.
+
+## Explicit non-goals for the first SaaS release
+
+- Teams/organisations and shared candidate data.
+- Auto-apply automation or scraping-proxy infrastructure.
+- A browser-extension marketplace release.
+- A billing-provider integration before commercial policy is decided.