Add activation batch attn and rename some attn logical names

[NuojCheng]

Description

This PR introduces a new logical name activation_batch_attn, which shares the same sharding rule with activation_batch for training. It (will) share different rules in vllm inference.

This PR also renames activation_attn_length to activation_length_attn to follow the tradition that compnent names should be at last.

Tests

• DS2 Performance test: https://diff.googleplex.com/#key=Qpj4ek6ZUz2u

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

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[NicoGrande]

LGTM, thank you for doing this!