[XPU] Add sage_attn backend

[xuechendi]

PLEASE FILL IN THE PR DESCRIPTION HERE ENSURING ALL CHECKLIST ITEMS (AT THE BOTTOM) HAVE BEEN CONSIDERED.

Purpose

Add sage_attn backend package

see 1.02% speedup per step; pure kernel see 1.26x gain so far.

Metric	SAGE_ATTN	FLASH_ATTN
Step 1	16.48s	16.66s
Step 2	12.94s	13.18s
Step 3	11.83s	12.09s
Step 4	11.30s	11.56s
Steps 5-17 (avg)	~10.6s	~10.88s
Step 18 (CPU offload spike)	14.01s	14.28s
Steps 19-40 (avg)	~10.6-10.7s	~10.87s
Total gen time	468.7s	480.0s

TODO:

1. work with cache_dit will hang

Test Plan

  export SYCL_UR_USE_LEVEL_ZERO_V2=0
  export ZE_AFFINITY_MASK=0,1,2,3
  ulimit -n 1048576
  VLLM_WORKER_MULTIPROC_METHOD=spawn \
  DIFFUSION_ATTENTION_BACKEND=SAGE_ATTN \
  python /workspace/vllm-omni/examples/offline_inference/text_to_video/text_to_video.py \
    --model Wan-AI/Wan2.2-T2V-A14B-Diffusers \
    --prompt "Two anthropomorphic cats in comfy boxing gear and bright gloves fight intensely on a spotlighted stage." \
    --height 480 --width 832 --num-frames 81 \
    --num-inference-steps 40 \
    --boundary-ratio 0.875 --flow-shift 5.0 --fps 16 \
    --tensor-parallel-size 4 --enable-cpu-offload \
    --vae-use-slicing --vae-use-tiling \
    --enable-diffusion-pipeline-profiler \
    --output t2v_480p_tp4_cpu_offload_sage.mp4

Test Result

t2v_480p_tp4_cpu_offload_sage.mp4

---

Essential Elements of an Effective PR Description Checklist

• The purpose of the PR, such as "Fix some issue (link existing issues this PR will resolve)".
• The test plan. Please provide the test scripts & test commands. Please state the reasons if your codes don't require additional test scripts. For test file guidelines, please check the test style doc
• The test results. Please paste the results comparison before and after, or the e2e results.
• (Optional) The necessary documentation update, such as updating supported_models.md and examples for a new model. Please run mkdocs serve to sync the documentation editions to ./docs.
• (Optional) Release notes update. If your change is user-facing, please update the release notes draft.

BEFORE SUBMITTING, PLEASE READ https://github.com/vllm-project/vllm-omni/blob/main/CONTRIBUTING.md (anything written below this line will be removed by GitHub Actions)