PyTorch: [Test] Enable test/profiler/test_profiler.py on XPU (#196466)

Executive Summary

Enables the device-generic tests in test/profiler/test_profiler.py on XPU. Most of the hardware classification this PR used to carry landed upstream in #198321 - #198324 . What is left, reviewable in order: Classify the inline-annotation tests as generic (they need no CUDA) and name the CUDA-gated metadata test class for its gate. Prime Kineto with the current accelerator instead of CUDA only.

Artificial Intelligence Architecture & Model Evaluation

From an artificial intelligence architecture, model weights governance, and inference efficiency perspective: - **Weights Accessibility & Sovereignty:** Evaluates whether weights are open for private self-hosting or locked behind centralized cloud APIs. - **Quantization & Edge Performance:** Kernel optimizations (4-bit/8-bit GGUF, AWQ, EXL2) allow high tokens-per-second on consumer GPUs and Apple Silicon. - **Reasoning & Architectural Scaling:** Scrutinizes mixture-of-experts (MoE), attention mechanisms, and fine-tuning datasets against open community benchmarks.

Impact on the Open Ecosystem

Protects developers and enterprises from proprietary black-box entrapment, fostering auditable, sovereign AI infrastructure.