Add @stdlib/python module for Python interoperability

[nbeerbower]

Introduces a new standard library module enabling Hemlock programs to use
Python libraries like Hugging Face Transformers, NumPy, PyTorch, etc.

Features:

• Initialize/finalize Python interpreter (py_init, py_finalize)
• Import Python modules (py_import, py_from_import, py_add_path)
• Call Python functions with args/kwargs (py_call, py_call_method, py_call_kw)
• Automatic type conversion (to_python, from_python)
• Object attribute access (py_get, py_set, py_has)
• Collection operations (py_getitem, py_setitem, py_len)
• Collection builders (py_list, py_dict, py_tuple, py_set_create)
• Type checking (py_type, py_is_int, py_is_string, etc.)
• Error handling (py_error_occurred, py_check_error, py_error_fetch)
• GIL management for async code (py_gil_acquire, py_with_gil)
• NumPy integration (numpy_from_buffer, numpy_to_buffer, numpy_shape)
• Convenience functions (py_eval, py_exec, py_repr)

Type conversion table:

• null <-> None
• bool <-> bool
• i8-i64, u8-u64 <-> int
• f32, f64 <-> float
• string <-> str
• array <-> list
• object <-> dict
• buffer <-> bytes

Example usage:
import { py_init, py_import, py_call, from_python } from "@stdlib/python";
py_init();
let math = py_import("math");
let result = py_call(py_get(math, "sqrt"), [16]);
print(from_python(result)); // 4.0
py_finalize();