[BUG] `load_model_trainer_states_from_checkpoint()` function is broken

Bug description

The transformers4rec.torch.trainer.load_model_trainer_states_from_checkpoint() function can only load t4rec_model_class.pkl, all the other loads are broken. The utils.serialization.load() function is called 4 times in the function, but only one of them successfully loads, the other throw errors. I believe this is because load() is expecting an _io.BufferedReader object (resulting from open()) instead of a path.

Correctly implemented:
https://github.com/NVIDIA-Merlin/Transformers4Rec/blob/5d59d14562cff667ce624f4a2a619a9803b7c4af/transformers4rec/torch/trainer.py#L744-L746

Incorrectly implemented:
https://github.com/NVIDIA-Merlin/Transformers4Rec/blob/5d59d14562cff667ce624f4a2a619a9803b7c4af/transformers4rec/torch/trainer.py#L753
https://github.com/NVIDIA-Merlin/Transformers4Rec/blob/5d59d14562cff667ce624f4a2a619a9803b7c4af/transformers4rec/torch/trainer.py#L757
https://github.com/NVIDIA-Merlin/Transformers4Rec/blob/5d59d14562cff667ce624f4a2a619a9803b7c4af/transformers4rec/torch/trainer.py#L766

Steps/Code to reproduce bug

Run the following code with a Model checkpoint at /checkpoint:

from transformers4rec import torch as tr

trainer = tr.trainer.Trainer(model=tr.model.base.Model(), args=None)
trainer.load_model_trainer_states_from_checkpoint('/checkpoint')

Expected behavior

The function call shouldn't be erroring.

Environment details

• Transformers4Rec version: 23.12.0
• Platform: Linux
• Python version: 3.10.12
• Huggingface Transformers version: 4.27.1
• PyTorch version (GPU?): 2.1.2 (no)

Start in transformers4rec/torch/trainer.py at load_model_trainer_states_from_checkpoint(), especially the working load call around lines 744-746 and the other calls around lines 753, 757, and 766. Reproduce with the provided checkpoint-loading example, then verify that all trainer states load without errors rather than only t4rec_model_class.pkl.