Update API to match latest llama.cpp version

[mamei16]

This PR updates llama_cpp.py so that it matches the llama.h API changes introduced in the commits:
ggml-org/llama.cpp@e0dbec0
ggml-org/llama.cpp@8fcb563
ggml-org/llama.cpp@00d5380
ggml-org/llama.cpp@dd373dd
ggml-org/llama.cpp@b3de7ca
ggml-org/llama.cpp@2c3f8b8
ggml-org/llama.cpp@e0e912f

I couldn't find any example on how to handle deprecated methods in this project, so I added a @deprecated decorator to the methods in question.

[mamei16]

I wasn't sure if c_void_p is the correct type to use here, so feel free to change it if there's a better alternative.

[jonathanreichhealthscope]

Looks great. Seems like there's a couple extra?

@ctypes_function(
    "llama_load_model_from_file",
    [ctypes.c_char_p, llama_model_params],
    llama_model_p_ctypes,
)
def llama_load_model_from_file(
    path_model: bytes, params: llama_model_params, /
) -> Optional[llama_model_p]:
    ...
    
 
# LLAMA_API void llama_model_free(struct llama_model * model);
@ctypes_function(
    "llama_model_free",
    [llama_model_p_ctypes],
    None,
)
def llama_model_free(model: llama_model_p, /):
    ...

[mamei16]

@jonathanreichhealthscope Thanks! What do you mean by that?

[jonathanreichhealthscope]

@jonathanreichhealthscope Thanks! What do you mean by that?

Looks like there's a couple of extra API changes, that's what I've put in my comment. For example, the 'llama_load_model_from_file' is now best practice to use, as the former naming convention I think is deprecated.