Generate a Tool Schema from a Python Function

LLM tool calling needs a JSON Schema for each function. Frameworks generate it automatically from type hints. Do the same with inspect.

Implement tool_schema(fn) returning:

{"name": fn.__name__, "description": <first line of the docstring, or "">,
 "parameters": {"type": "object", "properties": {...}, "required": [...]}}

• Map annotations: str → "string", int → "integer", float → "number", bool → "boolean", list[X] → {"type": "array", "items": <schema of X>}, dict → "object". A missing annotation gives {} (any type).
• Literal["a", "b"] → {"type": "string", "enum": ["a", "b"]}.
• Optional[X] (or X | None) → the schema of X, and the parameter is not required.
• Parameters without defaults are required (in signature order). Parameters with a default include "default" in their schema. Skip self, *args and **kwargs.