Add maestro framework

[x41lakazam]

Maestro is a Pytorch-based benchmarking framework for overlapping communication/compute operations in distributed environments.
This presents a more accurate way to measure performance of AI workloads compared to the micro-benchmarking which measures performance of standalone operations.
It allows a user to define a workload pattern, then benchmark performance of every running block.

For each block, Maestro reports average, minimum, maximum, and 99th percentile (P99) latencies across iterations, plus average bandwidth. P99 latency captures tail behavior that the average can hide — see docs/doc.md for the full output format.

[meta-cla]

Hi @x41lakazam!

Thank you for your pull request and welcome to our community.

Action Required

In order to merge any pull request (code, docs, etc.), we require contributors to sign our Contributor License Agreement, and we don't seem to have one on file for you.

Process

In order for us to review and merge your suggested changes, please sign at https://code.facebook.com/cla. If you are contributing on behalf of someone else (eg your employer), the individual CLA may not be sufficient and your employer may need to sign the corporate CLA.

Once the CLA is signed, our tooling will perform checks and validations. Afterwards, the pull request will be tagged with CLA signed. The tagging process may take up to 1 hour after signing. Please give it that time before contacting us about it.

If you have received this in error or have any questions, please contact us at cla@meta.com. Thanks!