Allow custom Langchain LLM object to be passed in `create_coding_agent`

[nb-programmer]

When solving a task or creating the agent directly, instead of using the openrouter instantiation, allow llm to be passed as an argument. If not, use openrouter as default.

agentic-python-coder/coder/src/agentic_python_coder/agent.py

Lines 32 to 44 in ba9c153

	def create_coding_agent(
	working_directory: str,
	system_prompt: Optional[str] = None,
	system_prompt_path: Optional[str] = None,
	model: str = None,
	project_prompt: Optional[str] = None,
	with_packages: Optional[List[str]] = None,
	task_content: Optional[str] = None,
	task_basename: Optional[str] = None,
	api_key: Optional[str] = None,
	todo: bool = False,
	verbose: bool = False,
	):

this way, we can pass an llm model ourselves. Example:

import agentic_python_coder as coder

from langchain.chat_models import init_chat_model

llm = init_chat_model("openai:gpt-4o-mini")

agent = coder.create_coding_agent(
    working_directory=working_dir,
    model=llm,  # Custom llm model from langchain
    task_content="...",
    task_basename=None,
    todo=True,
    verbose=True,
)