Draft PR to add GoodPut Measurement

[jiya-zhang]

Adding demo code as a skeleton to integrate GoodPut measurement into AXLearn.

To do's:

1. Consider localizing GoodPut measurement to within trainer.py
2. Make the measurement configurable
3. Automatically pick up run_name from job config
4. In order to obtain the GoodPut throughout a training's life time, consider repeatedly calling GoodPut Calculator and write the results to a persistent data store
5. Add tests in AXLearn
6. Test E2E for multislice jobs and larger single slice jobs
7. Test E2E for long-running jobs (8hr+)
8. Test for jobs running on v5e, v5p (so far, we only tested v4-8)