fix: apply unified APU name fallback to get_gpu_info_amd_detailed()

The detailed GPU function still used `or "AMD Radeon"` for unified APUs
while get_gpu_info_amd() was updated to use get_gpu_tier(). This caused
/api/gpu/detailed to return "AMD Radeon" while /gpu returned "Strix Halo 90+".

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: Tony363

dream-server/extensions/services/dashboard-api/gpu.py

@@ -503,7 +503,13 @@ def get_gpu_info_amd_detailed() -> Optional[list[IndividualGPU]]:
                 if power_str:
                     power_w = round(int(power_str) / 1e6, 1)
 
-            gpu_name = _read_sysfs(f"{base}/product_name") or "AMD Radeon"
+            gpu_name = _read_sysfs(f"{base}/product_name")
+            memory_type = "unified" if is_unified else "discrete"
+            if not gpu_name:
+                if is_unified:
+                    gpu_name = get_gpu_tier(mem_total / (1024**3), memory_type)
+                else:
+                    gpu_name = "AMD Radeon"
             card_id = base.split("/")[-2]  # "card0", "card1", …
             mem_used_mb = mem_used // (1024 * 1024)
             mem_total_mb = mem_total // (1024 * 1024)