feat(voice): DAVE E2E receive-side decrypt — unblocks Pas 12

Vendored fork: discord-ext-voice_recv 0.5.3a+echo.dave1

Patches the receive pipeline to handle Discord's mandatory DAVE E2E
encryption on voice gateway v=8. Without this, opus_decode raised
"corrupted stream" on every received packet in a DAVE-active room and
voice-to-voice never connected.

DAVE patch (vendor/discord-ext-voice-recv/reader.py):
- `_maybe_dave_decrypt(rtp_packet)`: gate mirrors discord.py 2.7.1
  `voice_state.can_encrypt`. Uses davey's `can_passthrough(user_id)` to
  branch — peers in passthrough send transport-only packets that pass
  through verbatim; peers in DAVE epoch go through `davey.decrypt`.
- Hooked in `callback()` between transport decrypt and feed_rtp;
  drops on decrypt failure without killing the reader thread.
- Bumps __version__ to '0.5.3a+echo.dave1' (PEP 440 local segment) so a
  contract test can fail fast on accidental upstream-sync overwrite.

Pipeline fixes uncovered while testing DAVE end-to-end:
- src/voice/pipeline.py: silero-vad v6+ requires exactly 512 samples per
  call at 16kHz; our 100ms window (1600 samples) was silently raising
  ValueError → VAD always returned False → STT never fired. Slice the
  window into 512-sample chunks. Bump whisper beam_size 1→5 and add a
  Romanian `initial_prompt` — transcriptions go from "Eco salt." gibberish
  to "Echo, salutare, te rog spune-mi cât este ora."
- src/voice/tts_stream.py: EchoStreamingAudioSource.read() returns a 20ms
  silence frame instead of b'' on empty queue. Empty return is treated
  by Discord as end-of-stream and kills the player, so any TTS pushed
  later would be silently discarded.
- src/adapters/discord_voice.py: actually attach EchoStreamingAudioSource
  to the voice client after the wakeup beep (chained via `after=`),
  which was missing entirely — TTS frames had no consumer.

Tests:
- tests/test_voice_recv_dave.py: 11 unit + callback integration tests
  covering bypass paths, can_passthrough gate, decrypt error handling.
- tests/test_voice_adapter_contract.py: +test_voice_recv_fork_version
  and +test_voice_connection_state_has_dave_attrs guards against
  upstream drift on either side.

Config:
- config.json: voice.allowed_user_ids whitelist for Marius's user id.

Status: voice-to-voice loop closes end-to-end (DAVE → VAD → Whisper →
Claude → Supertonic → audio out). Latency is ~8-13s per turn, which is
out of scope for this commit — see TODOS.md for the real-time UX
follow-up plan.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

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

src/adapters/discord_voice.py

@@ -21,6 +21,14 @@ from typing import Optional
 import discord
 from discord import app_commands
 
+# Optional DAVE dep (mandatory at runtime when discord.py 2.7.1 is paired with
+# Discord voice gateway v=8; tolerated missing in tests / dev environments).
+try:
+    import davey
+    _HAS_DAVE = True
+except ImportError:
+    _HAS_DAVE = False
+
 from src.config import Config
 from src.voice.pipeline import (
     VoiceSession,
@@ -28,7 +36,7 @@ from src.voice.pipeline import (
     _get_whisper_model,
     _get_silero_vad,
 )
-from src.voice.tts_stream import TTSQueue
+from src.voice.tts_stream import TTSQueue, EchoStreamingAudioSource
 from src.voice._discord_voice_adapter import connect_voice
 
 log = logging.getLogger("echo-core.discord.voice")
@@ -53,6 +61,11 @@ async def warmup_models() -> None:
     """
     global _voice_load_error
     try:
+        if not discord.opus.is_loaded():
+            discord.opus.load_opus("libopus.so.0")
+        if _HAS_DAVE:
+            log.info("DAVE protocol v%d available (davey %s)",
+                     davey.DAVE_PROTOCOL_VERSION, davey.__version__)
         await asyncio.to_thread(_get_whisper_model)
         await asyncio.to_thread(_get_silero_vad)
         log.info("Voice models warm")
@@ -167,11 +180,24 @@ def register(tree: app_commands.CommandTree, bot: discord.Client) -> app_command
             )
             return
         _voice_sessions[guild_id] = session
-        # Wake-up beep
+        # Start TTS streaming source for the entire session. Chain the
+        # wake-up beep via `after=` so streaming takes over when beep ends.
+        def _start_stream(error: Optional[Exception] = None) -> None:
+            if error is not None:
+                log.warning("Beep playback ended with error: %s", error)
+            try:
+                vc.play(EchoStreamingAudioSource(ttsq))
+                log.info("TTS streaming source attached")
+            except Exception:
+                log.exception("EchoStreamingAudioSource attach failed")
         try:
-            vc.play(discord.FFmpegPCMAudio("assets/voice/beep_200ms.wav"))
+            vc.play(
+                discord.FFmpegPCMAudio("assets/voice/beep_200ms.wav"),
+                after=_start_stream,
+            )
         except Exception:
-            log.warning("Beep playback skipped", exc_info=True)
+            log.warning("Beep playback skipped, starting stream directly", exc_info=True)
+            _start_stream()
         # Attach sink
         try:
             bot_user_id = int(bot.user.id) if bot.user is not None else 0