perf(voice): run STT+TTS on the GPU by default with CPU fallback

Both voice backends defaulted to CPU. Fix the "CUDA unavailable" gaps so
everything that benefits from the RTX 5050 uses it:

- MeloTTS venv had CPU-only torch (2.12.0+cpu) -> installed Blackwell-capable
  torch/torchaudio 2.11.0+cu128 (sm_120 verified with a real GPU matmul).
- faster-whisper CUDA loaded but transcribe() died with "libcublas.so.12 not
  found": installed nvidia-cublas-cu12 + nvidia-cudnn-cu12 into the whisper
  venv and inject those nvidia/*/lib dirs into the worker's LD_LIBRARY_PATH at
  spawn (the loader only honours it at exec).
- WSAI_WHISPER_DEVICE / WSAI_MELO_DEVICE now default to "auto": pick CUDA when
  present, else CPU, and each worker falls back to CPU if a CUDA load fails so
  the voice loop never dies on a GPU-less host.

Verified end-to-end through the real backend classes: both workers report
"ready on cuda"; steady-state STT ~170ms (was ~1350ms CPU), TTS ~4s first call
vs ~23s CPU. All 12 tests pass.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

wsai/backends/melo_worker.py

@@ -38,11 +38,33 @@ def _log(*a):
 
 def main() -> None:
     lang = "KR"
-    device = os.environ.get("WSAI_MELO_DEVICE", "cpu")  # "cpu" | "cuda" | "auto"
-    t0 = time.monotonic()
+    requested = os.environ.get("WSAI_MELO_DEVICE", "auto")  # cpu | cuda | auto
     from melo.api import TTS  # heavy import; only in the melo venv
 
-    tts = TTS(language=lang, device=device)
+    def _has_cuda() -> bool:
+        try:
+            import torch
+
+            return torch.cuda.is_available()
+        except Exception:
+            return False
+
+    device = requested
+    if requested == "auto":
+        device = "cuda" if _has_cuda() else "cpu"
+
+    t0 = time.monotonic()
+    try:
+        tts = TTS(language=lang, device=device)
+    except Exception as exc:
+        # CUDA picked but unusable (CPU-only torch, missing libs, OOM): fall back
+        # to CPU rather than leaving the whole voice loop dead.
+        if device == "cuda":
+            _log(f"[melo_worker] CUDA load failed ({exc}); falling back to CPU")
+            device = "cpu"
+            tts = TTS(language=lang, device=device)
+        else:
+            raise
     speaker_id = tts.hps.data.spk2id[lang]
     load_ms = int((time.monotonic() - t0) * 1000)
     _emit({"ready": True, "ms": load_ms, "device": device})