Update MLPerfCallback for v2.1

[hanlint]

What does this PR do?

Updates the MLPerfCallback by adding bert support, and updating the MLperf compliance checker to the v2.1 ruleset. Also fixes the tests to be passing.

Detailed fixes to MLPerfCallback:

• train_samples is now logged properly when using IterableDataset, which has a per-device length.
• the BERT benchmark logs time in units of samples, whereas resnet logs in epochs
• uploading the results files now occurs at the end of every eval step, and also when the callback is closed.
• adds option exit_at_target to exit training when target is met. Default: False

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)

[dskhudia]

LGTM! Thank you

[dakinggg]

LGTM, one comment because I just merged #1601

[eracah]

LGTM