docs: improve lora guide and demo notebook

[RexBearIU]

Description

This PR addresses and resolves the outstanding post-merge review comments left on PR #3970.

Specifically:

• Adds UV_TORCH_BACKEND="cpu" environment variable: Set inside the Colab setup code block in lora_llama3_demo.ipynb prior to invoking uv to ensure proper dependency resolution for CPU/TPU VM environments and avoid huge CUDA package downloads.
• Removes redundant PyTorch installation: Cleaned up the redundant subprocess pip call to install PyTorch prior to checkpoint conversion in lora_llama3_demo.ipynb, as PyTorch is already installed via the post-training dependencies target (tpu-post-train).
• Fixes Relative Code Links: Converted relative links pointing outside the docs/ folder (such as ../../src/maxtext/...) to absolute GitHub URLs in docs/guides/lora_model_bringup.md so that they work correctly on the ReadTheDocs built documentation.

REPORTS: #3970

Tests

• Validated notebook JSON syntax and schema correctness using python3 -m json.tool.
• Inspected markdown links inside lora_model_bringup.md to ensure they point to correct absolute URLs on GitHub.

Checklist

• I have performed a self-review of my code. For an optional AI review, add the gemini-review label.
• I have necessary comments in my code, particularly in hard-to-understand areas.
• I have run end-to-end tests tests and provided workload links above if applicable.
• I have made or will make corresponding changes to the doc if needed, including adding new documentation pages to the relevant Table of Contents (toctree directive) as explained in our documentation.