core[patch]: substantially speed up @deprecated (#29016)

Resolves https://github.com/langchain-ai/langchain/issues/26918

Unit tests don't raise any additional `LangChainDeprecationWarning`.
Would like guidance on how to test this more thoroughly if needed.

Note: speed up for `bind_tools` path is shown below. This is
**redundant** with the speedup in
https://github.com/langchain-ai/langchain/pull/29015. I include it for
demonstration purposes.

Before:

![Screenshot 2025-01-03 at 12 54
50 PM](https://github.com/user-attachments/assets/87f289eb-4cad-4304-85f7-5c58c59080f1)

After:

![Screenshot 2025-01-03 at 12 55
35 PM](https://github.com/user-attachments/assets/95ad0506-e1d1-4c5c-bb27-6a634d8810c9)

libs/core/langchain_core/_api/internal.py

@@ -14,10 +14,9 @@ def is_caller_internal(depth: int = 2) -> bool:
             frame = frame.f_back
             if frame is None:
                 return False
-        caller_module = inspect.getmodule(frame)
-        if caller_module is None:
-            return False
-        caller_module_name = caller_module.__name__
+        # Directly access the module name from the frame's global variables
+        module_globals = frame.f_globals
+        caller_module_name = module_globals.get("__name__", "")
         return caller_module_name.startswith("langchain")
     finally:
         del frame