feat: HA light output targets — cast LED colors to Home Assistant lights

New output target type `ha_light` that sends averaged LED colors to HA
light entities via WebSocket service calls (light.turn_on/turn_off):

Backend:
- HARuntime.call_service(): fire-and-forget WS service calls
- HALightOutputTarget: data model with light mappings, update rate, transition
- HALightTargetProcessor: processing loop with delta detection, rate limiting
- ProcessorManager.add_ha_light_target(): registration
- API schemas/routes updated for ha_light target type

Frontend:
- HA Light Targets section in Targets tab tree nav
- Modal editor: HA source picker, CSS source picker, light entity mappings
- Target cards with start/stop/clone/edit actions
- i18n keys for all new UI strings

server/src/wled_controller/api/routes/system.py

@@ -10,6 +10,8 @@ import sys
 from datetime import datetime, timezone
 from typing import Optional
 
+import os
+
 import psutil
 from fastapi import APIRouter, Depends, HTTPException, Query
 
@@ -52,8 +54,10 @@ from wled_controller.api.routes.system_settings import load_external_url  # noqa
 
 logger = get_logger(__name__)
 
-# Prime psutil CPU counter (first call always returns 0.0)
+# Prime psutil CPU counters (first call always returns 0.0)
 psutil.cpu_percent(interval=None)
+_process = psutil.Process(os.getpid())
+_process.cpu_percent(interval=None)  # prime process-level counter
 
 # GPU monitoring (initialized once in utils.gpu, shared with metrics_history)
 from wled_controller.utils.gpu import (  # noqa: E402
@@ -264,18 +268,36 @@ def get_system_performance(_: AuthRequired):
     """
     mem = psutil.virtual_memory()
 
+    # App-level metrics
+    proc_mem = _process.memory_info()
+    app_cpu = _process.cpu_percent(interval=None)
+    app_ram_mb = round(proc_mem.rss / 1024 / 1024, 1)
+
     gpu = None
     if _nvml_available:
         try:
             util = _nvml.nvmlDeviceGetUtilizationRates(_nvml_handle)
             mem_info = _nvml.nvmlDeviceGetMemoryInfo(_nvml_handle)
             temp = _nvml.nvmlDeviceGetTemperature(_nvml_handle, _nvml.NVML_TEMPERATURE_GPU)
+
+            # App GPU memory: sum memory used by this process on the GPU
+            app_gpu_mem: float | None = None
+            try:
+                pid = os.getpid()
+                for proc_info in _nvml.nvmlDeviceGetComputeRunningProcesses(_nvml_handle):
+                    if proc_info.pid == pid and proc_info.usedGpuMemory:
+                        app_gpu_mem = round(proc_info.usedGpuMemory / 1024 / 1024, 1)
+                        break
+            except Exception:
+                pass  # not all drivers support per-process queries
+
             gpu = GpuInfo(
                 name=_nvml.nvmlDeviceGetName(_nvml_handle),
                 utilization=float(util.gpu),
                 memory_used_mb=round(mem_info.used / 1024 / 1024, 1),
                 memory_total_mb=round(mem_info.total / 1024 / 1024, 1),
                 temperature_c=float(temp),
+                app_memory_mb=app_gpu_mem,
             )
         except Exception as e:
             logger.debug("NVML query failed: %s", e)
@@ -286,6 +308,8 @@ def get_system_performance(_: AuthRequired):
         ram_used_mb=round(mem.used / 1024 / 1024, 1),
         ram_total_mb=round(mem.total / 1024 / 1024, 1),
         ram_percent=mem.percent,
+        app_cpu_percent=app_cpu,
+        app_ram_mb=app_ram_mb,
         gpu=gpu,
         timestamp=datetime.now(timezone.utc),
     )