Metrics dedup for MLflow logger

[chenmoneygithub]

trafficstars

What does this PR do?

Context: MLflow has receiving lots of data... so we want to save some budget.

In this PR, if we detect that a metric has the same value as the previous step, we don't log it unless it meets one of the following 2 criteria:

• There has been log_duplicated_metric_every_n_steps skipped since last step logging, default to 100.
• There has been log_duplicated_metric_every_n_millis passed since last step logging, default to 600000 (10 minutes).

We put this exception for sampling purpose, otherwise it will be challenging to capture the duplicated values during sampling, despite the duplicated value occupies a significant portion of the metric history.

What issue(s) does this change relate to?

Before submitting

• [ ] Have you read the contributor guidelines?
• [ ] Is this change a documentation change or typo fix? If so, skip the rest of this checklist.
• [ ] Was this change discussed/approved in a GitHub issue first? It is much more likely to be merged if so.
• [ ] Did you update any related docs and document your change?
• [ ] Did you update any related tests and add any new tests related to your change? (see testing)
• [ ] Did you run the tests locally to make sure they pass?
• [ ] Did you run pre-commit on your change? (see the pre-commit section of prerequisites)