Add training and norm stat computation to module

[NigelNelson]

Hello,

To facilitate including openpi into various simulation frameworks as a supported model for training / inference, it would be very helpful if you could include the train.py and compute_norm_stats.py files in a python module within the repo. This way, external projects can pip install openpi as a dependency, and simply run:

from openpi import train, compute_norm_stats

args = parse_args()
# Get configuration using the provided config name
config = get_config(name=args.config, repo_id=args.repo_id, exp_name=args.exp_name)
# Ensure norm stats are computed
compute_norm_stats(config)
# Begin training
train.main(config)