Fix torch_logs recipe CPU fallback and device gate (#137285)

[yyouretoast]

Fixes pytorch/pytorch#137285.

Currently, the torch_logs tutorial/recipe renders as "Skipped" on the PyTorch website because the documentation pipeline builds on CPU-only CI environments where torch.cuda.is_available() is False.

This PR addresses the issue by:

1. Falling back to the CPU device when a CUDA device with compute capability >= 7.0 is unavailable.
2. Checking if torch.compile is supported dynamically in the build environment using a lightweight compilation test (torch.compile(lambda x: x + 1)).
3. Gracefully skipping execution using sys.exit(0) instead of throwing errors if the build environment lacks a C++ compiler.
4. Keeping the recipe's cells de-indented at the top-level so Sphinx-Gallery renders them as individual, step-by-step interactive code blocks on the website.

Testing

• Verified locally that the script exits with 0 when torch.compile is unsupported.
• Verified that fallback logic correctly selects the device and does not fail the execution flow.

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/tutorials/3924

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

⚠️ 4 Awaiting Approval

As of commit 038eaa5 with merge base 3cdec7b ():

AWAITING APPROVAL - The following workflows need approval before CI can run:

• Build tutorials (gh)
• Check Redirects for Deleted or Renamed Files (gh)
• link check on PR (gh)
• Lintrunner (gh)

This comment was automatically generated by Dr. CI and updates every 15 minutes.