feat: support for llama(.cpp) about mosec

[kemingy]

I found that there is a Python binding for llama.cpp: https://github.com/abetlen/llama-cpp-python

So this can be done in the following way:

Llama(Worker):
    def __init__(self_:
        self.model = llama_cpp.Llama(model_path="./models/llama-ggml.bin")

    def forward(self, data):
        return self.model(**data)

You can also use their low-level binding to create a pipeline if necessary.

[step21]

thanks!