refactor: update audio parameter handling in EdgeTTS and stubs

- Remove the unused `language`, `locale`, and `extra_parameters` parameters from the `EdgeTTS` function signature in `g4f/Provider/audio/EdgeTTS.py`.
- Update voice selection logic to check for `"locale"` and `"language"` keys in the `audio` dictionary, defaulting to `cls.default_locale` when neither is provided, and modify the error message accordingly.
- Refactor extraction of extra parameters by building a dict from the `audio` dictionary for keys `"rate"`, `"volume"`, and `"pitch"`.
- In `g4f/api/stubs.py`, remove the try/except block for importing `Annotated` and import `Messages` from `..typing` instead.
- Add an optional `audio: Optional[dict] = None` field to the `ImageGenerationConfig` model.

g4f/api/stubs.py

@@ -2,12 +2,8 @@ from __future__ import annotations
 
 from pydantic import BaseModel, Field, model_validator
 from typing import Union, Optional
-try:
-    from typing import Annotated
-except ImportError:
-    class Annotated:
-        pass
-from g4f.typing import Messages
+
+from ..typing import Messages
 
 class ChatCompletionsConfig(BaseModel):
     messages: Messages = Field(examples=[[{"role": "system", "content": ""}, {"role": "user", "content": ""}]])
@@ -69,6 +65,7 @@ class ImageGenerationConfig(BaseModel):
     n: Optional[int] = None
     negative_prompt: Optional[str] = None
     resolution: Optional[str] = None
+    audio: Optional[dict] = None
 
     @model_validator(mode='before')
     def parse_size(cls, values):