Add audio capture timing metrics to target pipeline

Instrument AudioCaptureStream with read/FFT timing and
AudioColorStripStream with render timing. Display audio-specific
timing segments (read/fft/render/send) in the target card
breakdown bar when an audio source is active.

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

server/src/wled_controller/core/audio/audio_capture.py

@@ -123,6 +123,9 @@ class AudioCaptureStream:
         self._smooth_spectrum_left = np.zeros(NUM_BANDS, dtype=np.float32)
         self._smooth_spectrum_right = np.zeros(NUM_BANDS, dtype=np.float32)
 
+        # Per-iteration timing (written by capture thread, read by consumers)
+        self._last_timing: dict = {}
+
     def start(self) -> None:
         if self._running:
             return
@@ -150,6 +153,10 @@ class AudioCaptureStream:
         with self._lock:
             return self._latest
 
+    def get_last_timing(self) -> dict:
+        """Return per-iteration timing from the capture loop (ms)."""
+        return dict(self._last_timing)
+
     def _capture_loop(self) -> None:
         try:
             import pyaudiowpatch as pyaudio
@@ -211,6 +218,7 @@ class AudioCaptureStream:
             spectrum_buf_right = np.zeros(NUM_BANDS, dtype=np.float32)
 
             while self._running:
+                t_read_start = time.perf_counter()
                 try:
                     raw_data = stream.read(self._chunk_size, exception_on_overflow=False)
                     data = np.frombuffer(raw_data, dtype=np.float32)
@@ -218,6 +226,7 @@ class AudioCaptureStream:
                     logger.warning(f"Audio read error: {e}")
                     time.sleep(0.05)
                     continue
+                t_read_end = time.perf_counter()
 
                 # Split channels and mix to mono
                 if channels > 1:
@@ -275,6 +284,12 @@ class AudioCaptureStream:
                         beat = True
                         beat_intensity = min(1.0, (ratio - 1.0) / 2.0)
 
+                t_fft_end = time.perf_counter()
+                self._last_timing = {
+                    "read_ms": (t_read_end - t_read_start) * 1000,
+                    "fft_ms": (t_fft_end - t_read_end) * 1000,
+                }
+
                 analysis = AudioAnalysis(
                     timestamp=time.perf_counter(),
                     rms=rms,