I add object::delete_dict() which changes Object's dict to None, and add PySetterValue::Delete match case in object::object_set_dict.
All things fine I think, but minor format issue remains.

    pub fn delete_dict(&self) -> Result<(), ()> {
        match self.instance_dict() {
            Some(_) => {
                unsafe {
                    let ptr = self as *const _ as *mut PyObject;
                    (*ptr).0.dict = None;
                }
                Ok(())
            }
            None => Err(()),
        }
    }

In this code, since Err can be returned, we should define custom Error type.
I made several options for it.

1. Return Ok(()) when it does not have dict already

• Can it satisfy normal convention? or I should raise error when the object to be deleted does not exist?

2. Make simple error type and return it.
3. Return Option not Result
4. Add something to delete_dict's parameter and return it.

I cannot decide which one is better. Can you suggest about it?