Remove FP16_Optimizer patch for DeepSpeed

Deepspeed already includes the same FusedAdam FP16_Optimizer originally from NVIDIA/apex here: https://github.com/deepspeedai/DeepSpeed/blob/master/deepspeed/runtime/fp16/fused_optimizer.py

Currently this line gives a warning saying:
```
/opt/conda/envs/py_3.10/lib/python3.10/site-packages/onnxruntime/training/optim/_modifier_registry.py:56: UserWarning: Skip modifying optimizer because of optimizer name not found in the registry: accelerate.utils.deepspeed.DeepSpeedOptimizerWrapper
```

i.e. the FP16 Optimizer from onnxruntime (https://github.com/microsoft/onnxruntime/blob/main/orttraining/orttraining/python/training/optim/fp16_optimizer.py) is not actually wrapping the DeepSpeed fused Adam optimizer anyway, so this line is redundant.

Author: Rohan138

optimum/onnxruntime/trainer.py

@@ -572,12 +572,6 @@ def _inner_training_loop(
                 )
             self.model = unwrap_model(model)
 
-        # ORT optimized FP16 optimizer for Deepspeed training
-        if self.is_deepspeed_enabled and args.fp16:
-            from onnxruntime.training.optim.fp16_optimizer import FP16_Optimizer
-
-            self.optimizer = FP16_Optimizer(self.optimizer)
-
         if self.is_fsdp_enabled:
             self.model = model