feat: add headers field to ModelConfigParam for custom LLM proxy auth

Adds optional `headers` field to `ModelConfigParam`, allowing users to
pass custom HTTP headers with every request to their model provider.
This aligns the Python SDK with the core TypeScript SDK which already
supports `headers` on `ClientOptions` and `ModelConfigObjectSchema`.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: shubh24

src/stagehand/types/model_config_param.py

@@ -2,6 +2,8 @@
 
 from __future__ import annotations
 
+from typing import Dict
+
 from typing_extensions import Literal, Required, Annotated, TypedDict
 
 from .._utils import PropertyInfo
@@ -19,5 +21,8 @@ class ModelConfigParam(TypedDict, total=False):
     base_url: Annotated[str, PropertyInfo(alias="baseURL")]
     """Base URL for the model provider"""
 
+    headers: Dict[str, str]
+    """Custom headers sent with every request to the model provider"""
+
     provider: Literal["openai", "anthropic", "google", "microsoft"]
     """AI provider for the model (or provide a baseURL endpoint instead)"""