feat: Add linear CE loss fusion for DPO

[pengdurice]

What does this PR do ?

Suport Linear CE Loss Fusion for DPO
On top of #2036 where Linear CE loss fusion support is added for SFT. This PR adds the support to DPO loss.

Issues

NA

Tests

1. Unit tests passed
2. Local tests (running the sh file) passed
3. A side by side loss curve comparison between baseline and experiment(linear ce loss fusion enabled)

Usage

# add the following to your dpo.yaml file.
megatron_cfg:
    enabled: true
    use_linear_ce_fusion_loss: true
    linear_ce_fusion_chunk_size: 256 # or other numbers 

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 run the unit tests and functional tests locally? Visit our Testing Guide for how to run tests
• Did you add or update any necessary documentation? Visit our Document Development Guide for how to write, build and test the docs.

Additional Information

• ...

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.