[onert/llm] Add 1st tokenizer for tinyllama

[glistening]

Caution

tinyllama is Maykeye/TinyLlama-v0

Currently GGMA generates a sequence of tokens, which is same to HF transformers generates.
However, it currently lack of a tokenizer support. Let's add tokenizer.

[glistening]

https://huggingface.co/Maykeye/TinyLLama-v0

I use tokenizer from open_llama_3b. However I had troubles with it locally(openlm-research/open_llama#69). I had no troubles on the cloud machine with preninstalled libraries.

[glistening]

I've done in #16056. I will make PR after translate support is done.

[glistening]

merged in #16270