Allow callers to pass pre-allocated moe_buf to avoid output copy

[tpopp]

Summary

• Add an optional moe_buf parameter to _moe_sorting_impl, moe_sorting, fused_moe, fused_moe_fake, and fused_moe_ functions
• When moe_buf is provided, the sorting kernel writes directly into the caller's buffer instead of allocating a new tensor, eliminating a redundant device-to-device copy on the output path

Test plan

• Verify existing MoE tests pass (no moe_buf → existing allocation behavior unchanged)
• Verify with moe_buf provided that the output is written in-place
• Benchmark to confirm reduced memory allocation overhead

Made with Cursor

[github-actions]

🏷️ CI Guide

Runs automatically on every PR:

• ✅ Pre-checks (submodule verification, code formatting)
• ✅ Aiter op tests (gfx942 + gfx950)
• ✅ Triton tests (only when aiter/ops/triton/** or related paths are changed)

Extended tests (opt-in via labels):

Label	Tests
ci:triton-355	Run Triton tests on MI355 in addition to MI325
ci:sglang	SGLang integration tests
ci:atom	ATOM benchmark (DeepSeek-R1 + GPT-OSS)
ci:vllm	vLLM benchmark
ci:all	All of the above

Add labels via the sidebar or gh pr edit 2663 --add-label <label>