feat: support EPLB for DeepSeek-V3.2 Python model executor.

[yingxudeng]

Add Expert Parallel Load Balancing (EPLB) support to the Python MoE path. The implementation uses slot-reuse mode where device_experts_num stays at num_local_experts to avoid exceeding the NPU fused GMM kernel groupList length limit. EPLB replaces cold expert weights in-place within fixed slots rather than allocating additional redundant slots.

Key changes:

• New xllm/python/layers/eplb.py with helper functions ported from C++
• PyCausalLM bridges prepare/start_transfer/update/last_ok to Python
• DeepseekV3MoE: log2phy_map remap in forward, dynamic lifecycle
• grouped_moe kernel: log2phy_map parameter for expert id remapping

Verified EP=2 (redundant=0) and EP=4 (redundant=1) outputs match baseline (EPLB off) token-for-token on 6-layer w8a8 model.

Description

Related Issues

Change Type

• Bug fix
• New feature
• Performance improvement
• Refactor
• Documentation
• Test
• Build or CI

Pull Request Checklist

Thank you for contributing to xLLM. Before requesting review, please make sure the following items are complete.

PR Title and Commit Messages

• The PR title and each commit message follow the xLLM commit format: <type>: <subject>.

Allowed types: feat, bugfix, docs, test, refactor, chore, style, revert, perf, model, build, release.
The subject should use clear English, start with a verb, include at least 4 words, and end with ..

Pre-commit Checks

• I have installed pre-commit by running pip install pre-commit or an equivalent command.
• I have installed the hooks with pre-commit install.
• I have run pre-commit run --all-files and fixed any reported issues.

If you are unsure how to set up pre-commit, see the pre-commit documentation.

Self Review

• I have self-reviewed the code according to .agents/skills/code-review/references/custom-code-style.md, especially code written or assisted by AI.
• I have rebased this PR onto the latest main branch.

Build and Test Coverage

• Tests have been added or updated as needed.
• CUDA: python setup.py build test has passed on a CUDA machine.
• NPU: python setup.py build test has passed on an NPU machine.
• MLU: python setup.py build test has passed on an MLU machine.

Reviewer Notes