feat: OpenVINO CPU MetricGAN mask; 100 ms hop

Per-worker OpenVINO CPU runs the MetricGAN BLSTM mask; SpeechBrain
keeps STFT/ISTFT. ENHANCE_INFER=auto with torch fallback. Hop is
0.1 s after isolated last_ms ~6 ms (model+ov+beamform).

run_worker.py

@@ -47,7 +47,8 @@ def main() -> int:
     ap.add_argument("--enhance-model",
                     default="speechbrain/metricgan-plus-voicebank")
     ap.add_argument("--enhance-chunk-s", type=float, default=1.0)
-    ap.add_argument("--enhance-hop-s", type=float, default=0.2)
+    ap.add_argument("--enhance-hop-s", type=float, default=0.1)
+    ap.add_argument("--enhance-infer", default="auto")
     ap.add_argument("--beamform-mode", default="auto")
     ap.add_argument("--torch-num-threads", type=int, default=1)
     ap.add_argument("--tdoa-every", type=int, default=5)
@@ -81,6 +82,7 @@ def main() -> int:
             beamform=args.beamform_mode,
             torch_num_threads=args.torch_num_threads,
             tdoa_every=args.tdoa_every,
+            enhance_infer=args.enhance_infer,
         )
         log.info("audio backend: %s", cleaner.backend)