[CI] Add llama case for profile test

[RUIJIEZHONG66166]

Add llama inference test for call number test when profiling

[pytorchxpubot]

@sys_pytorchxpubot triage result for run 15434458769

Triage bot UT analaysis result for reference only, please note unique error message only report once:

1. third_party.torch-xpu-ops.test.xpu.test_nn_xpu.TestNN test_LayerNorm_3d_no_affine_large_feature_cuda got failed with error message

 AssertionError: Tensor-likes are not close! 

Triage bot response:

{
  "similar_issue_id": 845,
  "similar_issue_state": "closed",
  "issue_owner": "daisyden",
  "issue_description": "The test TestNN.test_LayerNorm_3d_no_affine_large_feature_cuda failed with an AssertionError: Tensor-likes are not close! The error suggests a mismatch in tensor values between CUDA and XPU implementations. The failure is consistent and not random, indicating a potential discrepancy in the implementation or computation between the two devices.",
  "root_causes": [
    "Discrepancies in LayerNorm implementation between CUDA and XPU leading to tensor value mismatches.",
    "Potential differences in precision, kernel behavior, or synchronization issues between CUDA and XPU implementations."
  ],
  "suggested_solutions": [
    "Align the XPU LayerNorm implementation with CUDA to ensure consistent results.",
    "Investigate and correct any precision or kernel-specific issues causing the mismatch.",
    "If alignment is not feasible, consider temporarily skipping the test until the underlying issue is resolved."
  ]
}

[RUIJIEZHONG66166]

Ondemand test results: https://github.com/intel/torch-xpu-ops/actions/runs/16411411201/job/46370358970.

• Llama needs HF log in, but it works well on local.
• Other profile tests work well.