Add logic for sync distributed for TPUs in Step Result log

[kaushikb11]

🐛 Bug

Add logic for sync distributed for TPUs in Step Result log

https://github.com/PyTorchLightning/pytorch-lightning/blob/master/pytorch_lightning/core/step_result.py#L106

Environment

Note: Bugs with code are solved faster ! Colab Notebook should be made public !

• IDE: Please, use our python bug_report_model.py template.

• Colab Notebook: Please copy and paste the output from our environment collection script (or fill out the checklist below manually).

You can get the script and run it with:

wget https://raw.githubusercontent.com/PyTorchLightning/pytorch-lightning/master/tests/collect_env_details.py
# For security purposes, please check the contents of collect_env_details.py before running it.
python collect_env_details.py

• PyTorch Version (e.g., 1.0):
• OS (e.g., Linux):
• How you installed PyTorch (conda, pip, source):
• Build command you used (if compiling from source):
• Python version:
• CUDA/cuDNN version:
• GPU models and configuration:
• Any other relevant information:

Additional context