feat: pre-fetched caller context replaces tool-based patient lookups

- CallerContextService: fetches lead profile, appointments, call history,
  activities in parallel. Caches in Redis (5 min TTL). Renders as
  human-readable KB section — no UUIDs exposed to the LLM.
- Caller resolution controller: prewarms context cache on resolve
  (fire-and-forget) so the AI stream has a cache hit.
- AI chat stream: injects caller context into system prompt KB instead
  of raw Lead ID. LLM answers patient questions from context, no tool
  calls needed for current caller data.
- Eliminates UUID hallucination: LLM never sees leadId or patientId,
  can't pass wrong ID to wrong tool parameter.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

src/caller/caller-resolution.controller.ts

@@ -1,11 +1,15 @@
 import { Controller, Post, Body, Headers, HttpException, HttpStatus, Logger } from '@nestjs/common';
 import { CallerResolutionService } from './caller-resolution.service';
+import { CallerContextService } from './caller-context.service';
 
 @Controller('api/caller')
 export class CallerResolutionController {
     private readonly logger = new Logger(CallerResolutionController.name);
 
-    constructor(private readonly resolution: CallerResolutionService) {}
+    constructor(
+        private readonly resolution: CallerResolutionService,
+        private readonly callerContext: CallerContextService,
+    ) {}
 
     @Post('resolve')
     async resolve(
@@ -21,6 +25,12 @@ export class CallerResolutionController {
 
         this.logger.log(`[RESOLVE] Resolving caller: ${phone}`);
         const result = await this.resolution.resolve(phone, auth);
+
+        // Pre-warm caller context cache so the AI chat has it ready
+        if (result.leadId) {
+            this.callerContext.prewarm(result.leadId, result.patientId, auth);
+        }
+
         return result;
     }
 }