fix(tts): P1 TTS emotion instruction format per voice type + missing CN aliases (#1934)

fix(tts): P1 TTS emotion instruction format per voice type + missing CN aliases

- Add CN aliases 中性→neutral/伤心→sad/吃惊→surprised per 灵应 spec
- build_emotion_instruction(voice_id, enum): cloned voices use English "Speak in a {emotion} tone.", system voices with emotion Instruct (longanyang/longanhuan/longhuhu_v3) use strict Chinese format "你说话的情感是{emotion}。", default voice longxiaochun_v3 (no Instruct support) skips instruction entirely
- Expand preview emotion whitelist; 210 emotion tests passing, 15438 total unit tests passing
- E2E templates fix already in PR#1931 (0e5127df), no additional change needed

apps/api/app/api/routes/voice_clones.py

@@ -318,8 +318,12 @@ _ALLOWED_PREVIEW_EMOTIONS = {
     "悲伤",
     "愤怒",
     "惊奇",
+    "吃惊",
     "害怕",
     "讨厌",
+    # 灵应 P1 指定别名
+    "中性",
+    "伤心",
     "沉稳",
     "亲切",
 }
@@ -349,7 +353,7 @@ def get_voice_clone_preview(
     if emotion not in _ALLOWED_PREVIEW_EMOTIONS:
         raise HTTPException(
             status_code=status.HTTP_400_BAD_REQUEST,
-            detail=f"不支持的 emotion 值: {emotion}，可选: neutral/happy/sad/angry/surprised/fearful/disgusted 或中文 中立/开心/难过/生气/惊讶/恐惧/厌恶，或留空",
+            detail=f"不支持的 emotion 值: {emotion}，可选: neutral/happy/sad/angry/surprised/fearful/disgusted 或中文 中立/中性/开心/难过/伤心/生气/愤怒/惊讶/吃惊/恐惧/害怕/厌恶/讨厌 或留空",
         )
 
     use_case = GetVoiceCloneUseCase(repository)