Support both LangChain 0.3.x and 1.x via Python version markers

This commit enables LangChain 1.x support WITHOUT breaking changes by using
Python-version-conditional dependencies:

- Python 3.9 users: Continue using LangChain 0.3.x (no breaking change)
- Python 3.10+ users: Automatically get LangChain 1.x (new capability)

Changes:
- Add conditional dependency markers in pyproject.toml
- Regenerate poetry.lock with proper version markers
- Handle type compatibility between LangChain versions

This approach ensures CI testing works correctly:
- Python 3.9 tests use LangChain 0.3.x
- Python 3.10+ tests use LangChain 1.x
- Minimum version testing respects Python version constraints

Author: fede-kamel

libs/oci/langchain_oci/chat_models/oci_generative_ai.py

@@ -1358,7 +1358,7 @@ def with_structured_output(
                     key_name=tool_name, first_tool_only=True
                 )
         elif method == "json_mode":
-            llm = self.bind(response_format={"type": "JSON_OBJECT"})
+            llm = self.bind(response_format={"type": "JSON_OBJECT"})  # type: ignore[assignment]
             output_parser = (
                 PydanticOutputParser(pydantic_object=schema)
                 if is_pydantic_schema
@@ -1382,7 +1382,7 @@ def with_structured_output(
                 json_schema=response_json_schema
             )
 
-            llm = self.bind(response_format=response_format_obj)
+            llm = self.bind(response_format=response_format_obj)  # type: ignore[assignment]
             if is_pydantic_schema:
                 output_parser = PydanticOutputParser(pydantic_object=schema)
             else: