feat(tts): real Korean TTS via persistent MeloTTS worker

Adds a MeloTTS backend that runs the model in its own melo311 interpreter
as a long-lived worker (melo_worker.py), loaded once and fed synthesis
requests over a stdin/stdout JSON protocol. fd1 is split from fd2 in the
worker so MeloTTS's stdout progress chatter can't corrupt the protocol.
Each speak() writes a wav and hands the path to a pluggable sink (the
Discord voice step will swap in "play into the call"). factory wires
tts=melo; pipeline.aclose now also tears down the tts worker.

Verified (CPU): model load ~7.9s once, then a short reply synthesizes in
~0.86s (within the ~1s budget); wav is valid 44.1kHz PCM. GPU (cuda) is
selectable via WSAI_MELO_DEVICE for lower latency, pending GPU approval.
7 smoke tests still pass.

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

wsai/backends/melo_worker.py

@@ -0,0 +1,72 @@
+"""Persistent MeloTTS worker (Korean).
+
+MeloTTS lives in its own Python (melo311); loading the model takes seconds, so
+we load it ONCE here and then serve synthesis requests over stdin/stdout. This
+process is launched with the melo311 interpreter by wsai.backends.melo.MeloTTS.
+
+MeloTTS (and its deps) print progress straight to stdout, which would corrupt
+the JSON protocol. So on startup we split the streams: a private duplicate of
+the original stdout carries the protocol, and fd 1 is redirected to fd 2 so all
+library chatter lands on stderr instead.
+
+Protocol (one JSON object per line, on the protocol channel):
+    <- {"text": "...", "out": "/abs/path.wav", "speed": 1.3}
+    -> {"ok": true, "out": "/abs/path.wav", "ms": 123}
+    -> {"ok": false, "error": "..."}
+On startup, once the model is ready, it emits exactly one line:
+    -> {"ready": true, "ms": <load-ms>, "device": "cpu"}
+"""
+
+import json
+import os
+import sys
+import time
+
+# Split protocol from library noise BEFORE importing anything heavy.
+_proto = os.fdopen(os.dup(1), "w", buffering=1)  # private copy of real stdout
+os.dup2(2, 1)  # fd1 -> stderr, so stray library prints don't hit the protocol
+
+
+def _emit(obj: dict) -> None:
+    _proto.write(json.dumps(obj) + "\n")
+    _proto.flush()
+
+
+def _log(*a):
+    print(*a, file=sys.stderr, flush=True)
+
+
+def main() -> None:
+    lang = "KR"
+    device = os.environ.get("WSAI_MELO_DEVICE", "cpu")  # "cpu" | "cuda" | "auto"
+    t0 = time.monotonic()
+    from melo.api import TTS  # heavy import; only in the melo venv
+
+    tts = TTS(language=lang, device=device)
+    speaker_id = tts.hps.data.spk2id[lang]
+    load_ms = int((time.monotonic() - t0) * 1000)
+    _emit({"ready": True, "ms": load_ms, "device": device})
+    _log(f"[melo_worker] model ready in {load_ms} ms on {device}")
+
+    for line in sys.stdin:
+        line = line.strip()
+        if not line:
+            continue
+        try:
+            req = json.loads(line)
+            text = req["text"]
+            out = req["out"]
+            speed = float(req.get("speed", 1.0))
+            if out.startswith("/tmp") or out.startswith("/dev/shm"):
+                raise ValueError(f"refusing RAM-backed tmpfs path: {out}")
+            s = time.monotonic()
+            tts.tts_to_file(text, speaker_id, out, speed=speed)
+            ms = int((time.monotonic() - s) * 1000)
+            _emit({"ok": True, "out": out, "ms": ms})
+        except Exception as exc:  # keep the worker alive across bad requests
+            _emit({"ok": False, "error": f"{type(exc).__name__}: {exc}"})
+            _log(f"[melo_worker] error: {exc}")
+
+
+if __name__ == "__main__":
+    main()