Add section: GPU computing

[icaoberg]

Add a new section to the Sphinx documentation titled “GPU computing”. This section should show users how to utilize GPUs for compute workloads on the Lane cluster, including:

• How to request or select a GPU for computing.
• How to load/select CUDA.
• How to use TensorFlow with GPU support.

The section is intended for users familiar with basic cluster usage but new to GPU workflows.

Scope / Details

• Create a new Sphinx section (e.g., under Running Jobs or Advanced Usage) named GPU computing.
• Document the following:
  • How to check GPU availability on the Lane cluster.
  • Example job submission script requesting a GPU (scheduler directives).
  • Steps for loading CUDA modules or configuring the CUDA environment.
  • Instructions for setting up TensorFlow with GPU support (module, Conda environment, or virtualenv).
  • A simple TensorFlow GPU test snippet (e.g., listing visible GPUs).
• Optional but recommended: a troubleshooting section for common issues such as missing CUDA libraries or TensorFlow–CUDA version mismatches.

Acceptance Criteria

• A new GPU computing section is present in the generated Sphinx documentation.
• The section clearly explains:
  • How to request/select a GPU when submitting jobs on the Lane cluster.
  • How to load and use CUDA.
  • How to run TensorFlow with GPU support, including at least one example.
• Commands and code blocks render correctly in Sphinx.
• Documentation builds cleanly with no additional warnings.