add guard for self.weight.quant_state attribute

[winglian]

When saving qlora checkpoints with latest bnb, I get the error

stderr: [rank0]:   File "/home/wing/.venvs/axolotl/lib/python3.11/site-packages/transformers/trainer.py", line 3325, in _save_checkpoint
stderr: [rank0]:     self.save_model(output_dir, _internal_call=True)
stderr: [rank0]:   File "/home/wing/.venvs/axolotl/lib/python3.11/site-packages/transformers/trainer.py", line 4207, in save_model
stderr: [rank0]:     state_dict = self.accelerator.get_state_dict(self.model)
stderr: [rank0]:                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
stderr: [rank0]:   File "/home/wing/.venvs/axolotl/lib/python3.11/site-packages/accelerate/accelerator.py", line 4057, in get_state_dict
stderr: [rank0]:     state_dict = model.state_dict()                                                                          
stderr: [rank0]:                  ^^^^^^^^^^^^^^^^^^
stderr: [rank0]:   File "/home/wing/.venvs/axolotl/lib/python3.11/site-packages/torch/nn/modules/module.py", line 2260, in state_dict
stderr: [rank0]:     module.state_dict(  
stderr: [rank0]:   File "/home/wing/.venvs/axolotl/lib/python3.11/site-packages/torch/nn/modules/module.py", line 2260, in state_dict
stderr: [rank0]:     module.state_dict(
stderr: [rank0]:   File "/home/wing/.venvs/axolotl/lib/python3.11/site-packages/torch/nn/modules/module.py", line 2260, in state_dict
stderr: [rank0]:     module.state_dict(
stderr: [rank0]:   [Previous line repeated 7 more times]                                                                      
stderr: [rank0]:   File "/home/wing/.venvs/axolotl/lib/python3.11/site-packages/torch/nn/modules/module.py", line 2257, in state_dict
stderr: [rank0]:     self._save_to_state_dict(destination, prefix, keep_vars)
stderr: [rank0]:   File "/home/wing/.venvs/axolotl/lib/python3.11/site-packages/bitsandbytes/nn/modules.py", line 518, in _save_to_state_dict                                                                                                                
stderr: [rank0]:     if getattr(self.weight.quant_state, "packing_format_for_cpu", False):                                                                                                                                                                   
stderr: [rank0]:                ^^^^^^^^^^^^^^^^^^^^^^^                                                                                                                                                                                                      
stderr: [rank0]: AttributeError: 'Parameter' object has no attribute 'quant_state'

This was a regression from #1804 where it does not check quant_state is a valid attribute. If you look just a few lines below, we are checking for that attribute there too.
https://github.com/bitsandbytes-foundation/bitsandbytes/blob/main/bitsandbytes/nn/modules.py#L523