Fix CSP blocking IPC/assets + fix pyannote AudioDecoder crash

CSP: Add connect-src for ipc.localhost and asset.localhost so Tauri IPC
commands and local file loading (waveform, audio playback) work.

pyannote: Block torchcodec in sys.modules at startup so pyannote.audio
falls back to torchaudio for audio decoding. pyannote has a bug where
it uses AudioDecoder unconditionally even when torchcodec import fails.

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

python/voice_to_notes/main.py

@@ -5,6 +5,13 @@ from __future__ import annotations
 import signal
 import sys
 
+# Block torchcodec before anything imports it. pyannote.audio has a bug where
+# it conditionally imports AudioDecoder from torchcodec but uses it
+# unconditionally. Making torchcodec fully unimportable forces the torchaudio
+# fallback path.
+sys.modules["torchcodec"] = None  # type: ignore[assignment]
+sys.modules["torchcodec.decoders"] = None  # type: ignore[assignment]
+
 # CRITICAL: Capture real stdout for IPC *before* importing any ML libraries
 # that might print to stdout and corrupt the JSON-line protocol.
 from voice_to_notes.ipc.protocol import init_ipc