Loss Weighting Overridden in IBP Training

[GiulioZizzo]

Describe the bug
If the user chooses not to use a scheduler for the loss weighting, and wants to use a fixed weighting, their supplied value is overridden and a default one is used instead.

To Reproduce
Steps to reproduce the behavior:

1. Go to ibp_certified_trainer_pytorch.py
2. Scroll down to line 304
3. loss_weighting_k = 0.1 should be loss_weighting_k = self.loss_weighting

System information (please complete the following information):

• OS: Mac OS
• Python version: Python 3.8
• ART version or commit number: 1.14.0
• PyTorch 1.13.1