seperate quantize and export_to_edge in builder (#3613)

Summary:
Pull Request resolved: https://github.com/pytorch/executorch/pull/3613

Currently export_to_edge includes both applying quantizer and run to_edge, separate them so I can call quantize only in the eval_llama.py

Reviewed By: Jack-Khuu, larryliu0820

Differential Revision: D57367832

fbshipit-source-id: 04d225df5403657cd86726af8cffb77c7c41147e