[JAX] Support Ring Attention (Context Parallelism)

[mingxu1067]

Description

Please include a brief summary of the changes, relevant motivation and context.

• Adding utils of sharding to retrieve the given mesh information.
• Adding support for Ring Attention on top of Fused Attention.
• Adding workload balance of causal masking utils.
• Adding corresponding unit-tests.

Type of change

• Documentation change (change only to the documentation, either a fix or a new content)
• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Infra/Build change
• Code refractor

Changes

Please list the changes introduced in this PR:

• Adding utils of sharding to retrieve the given mesh information.
• Adding FWD and BWD primitives of Ring Attention.
• Adding a workload balance of causal masking function and its status tracker.
• Adding corresponding unit-tests.

Checklist:

• I have read and followed the contributing guidelines
• The functionality is complete
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes