feat: DPO support for global padding of seq_len to a multiple

[terrykong]

What does this PR do ?

• adds pad_to_multiple_of for DPO which is a requirement for sequence parallel (which is required for moe models w/ TP)
  • the argument pad_length_to_multiple_of will pad all minibatches to the same length if >0. If ==0, the behavior is the same as before.

Needed for:

• sequence parallel
• mamba

Rebase stack

• feat: DPO support for global padding of seq_len to a multiple #386
• feat: adds helper script for running & summarizing functional tests #387
• chore: adds missing license header on tests and corrects file modes #388
• feat: support for mcore optimizer (to enable MoE) #380

Changelog

• Please update the CHANGELOG.md under next version with high level changes in this PR.

Usage

• You can potentially add a usage example below

# Add a code snippet demonstrating how to use this 

Before your PR is "Ready for review"

Pre checks:

• Make sure you read and followed Contributor guidelines
• Did you write any new necessary tests?
• Did you add or update any necessary documentation? Make sure to also update the NeMo Framework User Guide which contains the tutorials

Checklist when contributing a new algorithm

• Does the trainer resume and restore model state all states?
• Does the trainer support all parallelism techniques(PP, TP, DP)?
• Does the trainer support max_steps=-1 and validation?
• Does the trainer only call APIs defined in alignable_interface.py?
• Does the trainer have proper logging?

Additional Information

• Related to # (issue)

[trias702]

LGTM, just a few minor questions. Also, to confirm, this PR just lets use pad DPO sequences to a certain multiple, this is not the PR which adds support to DPO for sequence parallelism, correct?

[trias702]

Just one added question

[trias702]

Sound