fix(external-llm): align tier defaults with industry benchmarks

- frontier: gpt-5.2 (GPT)
- mid-tier: claude-sonnet-4.5 (Claude)
- lightweight: claude-haiku-4.5 (Claude)

Prioritizes correctness over speed, aligned with MMLU/GPQA/Arena rankings.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

state/model-policy.json

@@ -184,8 +184,8 @@
   },
   "tier_to_external_map": {
     "frontier": "github-copilot/gpt-5.2",
-    "mid-tier": "github-copilot/gpt-5-mini",
-    "lightweight": "opencode/gpt-5-nano"
+    "mid-tier": "github-copilot/claude-sonnet-4.5",
+    "lightweight": "github-copilot/claude-haiku-4.5"
   },
   "task_routing": {
     "reasoning": "github-copilot/gpt-5.2",