Qualcomm AI Engine Direct - Set llama io as quantized tensor (#5383)

* Qualcomm AI Engine Direct - Add llama io be quantized

- Add general function to tag io obtain/genetate quantized tensor
- Add quantizing io function to llama2.py

* [Fix lint]

---------

Co-authored-by: Joey Tsai <chunit@qti.qualcomm.com>