Disable fused causal attention (#14732)
There is accuracy regression in GPT-2 model. Top1 match rate (vs PyTorch
model) drops about 1%. The cause is the fused causal attention uses fp16
accumulation. Disable it by default and add an environment variable
ORT_ENABLE_FUSED_CAUSAL_ATTENTION=1 to turn on it manually.
It also updated the GPT-2 parity test script to generate left side
padding to reflect the actual usage.
To test:
```
python -m onnxruntime.transformers.models.gpt2.convert_to_onnx -m gpt2 --output gpt2.onnx -o -p fp16 --use_gpu
```
The top1-match-rate in the output is on-par with ORT 1.13.1.
