convrot: add a hook to run post-quant custom setup when needed

[bghira]

This pull request introduces a new extensibility hook to support additional setup steps after manual quantization of models. The main changes involve adding a post_quantization_setup method to the base model class and invoking this hook in the training process after quantization is applied.

Model extensibility improvements:

• Added a post_quantization_setup method to the base model class in common.py as a hook for model families that require extra setup after manual quantization.

Training process integration:

• Updated the training pipeline in trainer.py to call post_quantization_setup on the model after quantization is performed, ensuring any necessary post-quantization steps are executed.