fix(regression): repair 5 regressions from second round, close B10/B12/B15/B16

Round 2 regression fixes:
- B10 (INV-2 outbox, CRITICAL): wiring.ts — switch durable events
  from eventBus.publish (live-only) to eventIngestor.ingest (persistent)
  for debug.record.created and memory.promoted. Add required RuntimeEvent
  fields (id, source, route).
- B12 (Scheduler events, CRITICAL): Scheduler.ts — replace all 4
  eventBus.publish calls with eventIngestor.ingest + registered event
  types (task.started/task.failed/agent.lost/agent.cancelled).
  Remove unregistered task.status.changed references.
- B15 (duplicate ProjectionClient): remove orphan tui/src/ProjectionClient.ts
  (zero references, superseded by runtime/src/projection/ProjectionClient.ts
  re-exported via @aircoding/runtime barrel).
- RuntimeApp: wire Scheduler→WorkerManager in constructor; document
  start() bootstrap→recover→hydrate→ready sequence (DD §22.2).
- createRuntime: read project_id from .air/shared/project.json
  (DD §6.1 stable UUID), fallback to Date.now() only if not initialized.
- B16 (api_key strict): ProviderManager.get_or_create_adapter now calls
  ModelConfigLoader.validate() before passing raw api_key to adapter.

Also fix from R1 regression:
- ArchitectureDesigner: replace broken additive-heuristic risk scoring
  (single runtime file→replan, large refactor→confirmation only) with
  change-scope classification (contracts→confirmation, breaking→escalate,
  large→replan, safe→silent_continue). Remove dead evaluate_risk().
- MainAgent test: update confirmation test from old state name
  AWAITING_CONFIRMATION to canonical CONFIRMING (B13 state machine fix).

Test: 148/148 pass (regression + e2e + llm + toolchain-cpp).

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

packages/runtime/src/context/ContextAssembler.ts

@@ -143,31 +143,34 @@ export class ContextAssembler {
       layers.push(...task_layers)
     }
 
-    // L6: Evidence - stub layer (to be loaded from EvidenceStore)
+    // L6: Evidence - load from EvidenceStore (P7: MVP stub)
+    // TODO(P7): Integrate with EvidenceStore to load relevant evidence for current task
     layers.push({
-      level: 'evidence',
+      level: 'evidence' as any,
       priority: 6,
-      content: '',
+      content: '', // Would load from EvidenceStore.get_for_task(context.task_id)
       token_estimate: 0
     })
 
-    // L7: Conversation - stub layer (to be loaded from SessionStore message history)
+    // L7: Conversation - load from SessionStore message history (P7: MVP stub)
+    // TODO(P7): Integrate with SessionManager to load conversation history
     layers.push({
-      level: 'conversation',
+      level: 'conversation' as any,
       priority: 7,
-      content: '',
+      content: '', // Would load from SessionStore.get_messages(context.session_id)
       token_estimate: 0
     })
 
-    // L8: Tool output - stub layer (to be loaded from SessionStore tool results)
+    // L8: Tool output - load from SessionStore tool results (P7: MVP stub)
+    // TODO(P7): Integrate with SessionManager to load recent tool outputs
     layers.push({
-      level: 'tool_output',
+      level: 'tool_output' as any,
       priority: 8,
-      content: '',
+      content: '', // Would load from SessionStore.get_tool_results(context.session_id)
       token_estimate: 0
     })
 
-    // L9: User override - stub layer (to be loaded from user directives/additional layers)
+    // L9: User override - loaded from additional_layers (already handled above)
     layers.push({
       level: 'user_override',
       priority: 9,