feat: Variet Engine v1.0 + 5-model tuning complete

Phase 01 (LLM Tuning):
- Gemma4 26B: 74.65 t/s (fast)
- Qwen 35B: 61.62 t/s (balanced)
- Gemma4 31B: 16.0 t/s (deep-coder)
- Qwen 27B: 16.7 t/s (deep-logic)
- Qwen 122B: 8.95 t/s (ultra, GPU 1 only)

Phase 02 (API Engine):
- FastAPI reverse proxy on port 8000
- /engine/switch hot-swap with 503 protection
- config/engine_models.json as single source of truth
- Replaced 4 individual .bat files with unified engine

File cleanup:
- scripts/ 85 files -> 9 + _archive/
- Root .bat files -> _archive/

scripts/_archive/results/122b_optimization_results.json

@@ -0,0 +1,43 @@
+[
+  {
+    "name": "ngl=999 + expert CPU + no-mmap",
+    "ngl": 999,
+    "avg_tps": 4.8,
+    "best_tps": 4.84,
+    "vram_gpu0": 4225,
+    "vram_gpu1": 3779,
+    "vram_total": 8004,
+    "pcie": "3, 4\r | 4, 16",
+    "status": "OK"
+  },
+  {
+    "name": "ngl=10 (pure, no expert override)",
+    "ngl": 10,
+    "avg_tps": 2.52,
+    "best_tps": 2.56,
+    "vram_gpu0": 10309,
+    "vram_gpu1": 5871,
+    "vram_total": 16180,
+    "pcie": "1, 4\r | 1, 16",
+    "status": "OK"
+  },
+  {
+    "name": "ngl=12 (pure)",
+    "ngl": 12,
+    "avg_tps": 2.86,
+    "best_tps": 2.86,
+    "vram_gpu0": 11807,
+    "vram_gpu1": 7377,
+    "vram_total": 19184,
+    "pcie": "2, 4\r | 2, 16",
+    "status": "OK"
+  },
+  {
+    "name": "ngl=14 (pure)",
+    "status": "BOOT_FAIL"
+  },
+  {
+    "name": "ngl=999 + upper expert CPU (blk 32-63)",
+    "status": "BOOT_FAIL"
+  }
+]