💥 Remove preload_models because it's useless

- We always load the model into memory during initialization. The option to not do it is not yet implemented and I don't consider it to be my top priority for v1.0.0

conf.CN.yaml

@@ -8,7 +8,6 @@ system_config:
   conf_version: "v1.0.0-alpha.1" # 配置文件版本
   host: "0.0.0.0" # 服务器监听的地址，"0.0.0.0" 表示监听所有网络接口；如果需要安全，可以使用 "127.0.0.1"（仅本地访问）
   port: 12393 # 服务器监听的端口
-  preload_models: True # 如果为 True，则在服务器启动时初始化 ASR（语音识别）和 TTS（语音合成）模型并将其保留在内存中
   config_alts_dir: "characters" # 用于存放替代配置的目录
   live2d_expression_prompt: "live2d_expression_prompt" # 将追加到系统提示末尾，让 LLM（大型语言模型）包含控制面部表情的关键字。支持的关键字将自动加载到 `[<insert_emomap_keys>]` 的位置。
 

conf.yaml

@@ -3,8 +3,6 @@ system_config:
   conf_version: "v1.0.0-alpha.1"
   host: "0.0.0.0" # use 127.0.0.1 if you want security
   port: 12393
-  # If true, ASR and TTS will be initialized when server starts and kept in memory
-  preload_models: True
   # New setting for alternative configurations
   config_alts_dir: "characters"
   # This will be appended to the end of system prompt to let LLM include keywords to control facial expressions.

src/open_llm_vtuber/config_manager/system.py

@@ -9,7 +9,6 @@ class SystemConfig(I18nMixin):
     conf_version: str = Field(..., alias="conf_version")
     host: str = Field(..., alias="host") 
     port: int = Field(..., alias="port")
-    preload_models: bool = Field(..., alias="preload_models")
     config_alts_dir: str = Field(..., alias="config_alts_dir")
     live2d_expression_prompt: str = Field(..., alias="live2d_expression_prompt")
 
@@ -26,10 +25,6 @@ class SystemConfig(I18nMixin):
             en="Server port number",
             zh="服务器端口号"
         ),
-        "preload_models": Description(
-            en="Whether to preload ASR and TTS models on startup",
-            zh="是否在服务器启动时预加载 ASR 和 TTS 模型"
-        ),
         "config_alts_dir": Description(
             en="Directory for alternative configurations",
             zh="备用配置目录"