Add support of Grain input pipeline for DPO.

[igorts-git]

Description

Add support of Grain input reading when using the new Tunix-based DPO/ORPO.

BUGS: b/485626968

Tests

CI tests.
Ran DPO/ORPO while reading using Grain from parquet files.

Checklist

Before submitting this PR, please make sure (put X in square brackets):

• I have performed a self-review of my code. For an optional AI review, add the gemini-review label.
• I have necessary comments in my code, particularly in hard-to-understand areas.
• I have run end-to-end tests tests and provided workload links above if applicable.
• I have made or will make corresponding changes to the doc if needed, including adding new documentation pages to the relevant Table of Contents (toctree directive) as explained in our documentation.

[codecov]

Codecov Report

❌ Patch coverage is 17.64706% with 28 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
...rc/maxtext/input_pipeline/grain_data_processing.py	12.50%	7 Missing ⚠️
src/maxtext/input_pipeline/tfds_data_processing.py	0.00%	6 Missing ⚠️
src/maxtext/trainers/pre_train/train.py	40.00%	6 Missing ⚠️
src/maxtext/utils/train_utils.py	0.00%	2 Missing and 1 partial ⚠️
src/maxtext/common/metric_logger.py	0.00%	2 Missing ⚠️
src/maxtext/common/train_state_nnx.py	33.33%	2 Missing ⚠️
src/maxtext/utils/gradient_accumulation.py	0.00%	2 Missing ⚠️

📢 Thoughts on this report? Let us know!