wrap llama_batch_get_one

[tinglou]

wrap llama_cpp_sys_2::llama_batch_get_one and safer drop

[MarcusDunn]

Why does this need to be mut? If it doesn't, can you make a PR to llama-cpp instead of casting? If it does need to be mut, that should be reflected in the signature.

Also what are the implications of an empty slice here?

[tinglou]

#[repr(C)]
#[derive(Debug, Copy, Clone, PartialEq)]
pub struct llama_batch {
    pub n_tokens: i32,
    pub token: *mut llama_token,
    pub embd: *mut f32,
    pub pos: *mut llama_pos,
    pub n_seq_id: *mut i32,
    pub seq_id: *mut *mut llama_seq_id,
    pub logits: *mut i8,
    pub all_pos_0: llama_pos,
    pub all_pos_1: llama_pos,
    pub all_seq_id: llama_seq_id,
}

The orignal field token is mut.

Yes, the slice should not be empty.