feat(voice): real TTS fallback chain and whisper-compatible STT

- TTS providers: cloud (OpenAI-compatible /audio/speech, e.g. Bailian
  compatible mode with cosyvoice), local self-hosted HTTP server,
  edge-tts CLI, espeak CLI, silent — ordered by TTS_PROVIDER list,
  skipping providers whose binary/config is missing
- STT: whisper-compatible /audio/transcriptions endpoint (cloud or
  self-hosted faster-whisper), replacing the empty dashscope stub
- config: TTS_CLOUD_URL/KEY/MODEL/VOICE, TTS_LOCAL_URL, STT_API_URL/
  KEY/MODEL; cloud keys fall back to DASHSCOPE_API_KEY
- gateway /voice/tts now returns real audio/mpeg bytes instead of a
  byte count
- tests: cloud/local/whisper round-trips via httptest, chain skip
  behavior, fallback order; kept and adapted prior service tests

backend/internal/config/loader.go

@@ -93,7 +93,15 @@ func (c *Config) mergeEnv() {
 	// Voice
 	c.DashScopeAPIKey = getEnv("DASHSCOPE_API_KEY", c.DashScopeAPIKey)
 	c.TTSProvider = getEnv("TTS_PROVIDER", c.TTSProvider)
+	c.TTSCloudURL = getEnv("TTS_CLOUD_URL", c.TTSCloudURL)
+	c.TTSCloudKey = getEnv("TTS_CLOUD_KEY", c.TTSCloudKey)
+	c.TTSCloudModel = getEnv("TTS_CLOUD_MODEL", c.TTSCloudModel)
+	c.TTSCloudVoice = getEnv("TTS_CLOUD_VOICE", c.TTSCloudVoice)
+	c.TTSLocalURL = getEnv("TTS_LOCAL_URL", c.TTSLocalURL)
 	c.ASRProvider = getEnv("ASR_PROVIDER", c.ASRProvider)
+	c.STTApiURL = getEnv("STT_API_URL", c.STTApiURL)
+	c.STTApiKey = getEnv("STT_API_KEY", c.STTApiKey)
+	c.STTModel = getEnv("STT_MODEL", c.STTModel)
 
 	// Auth
 	c.JWTSecret = getEnv("JWT_SECRET", c.JWTSecret)