fix(cookbook): emit HIP_VISIBLE_DEVICES for ROCm serves (#4950)

[Elshayib]

Summary

Serve runner scripts choose HIP_VISIBLE_DEVICES when launch env or command indicates ROCm instead of always exporting CUDA_VISIBLE_DEVICES.

Target branch

• This PR targets dev, not main. All PRs land in dev; main is curated by the maintainer at each release. If your PR is on main by accident, click "Edit" on this PR and change the base.

Linked Issue

Fixes #4950

Type of Change

• Bug fix (non-breaking — fixes a confirmed issue)

Checklist

• I searched open issues and open PRs — this is not a duplicate.
• This PR targets dev
• My changes are limited to the scope described above — no unrelated refactors or whitespace changes mixed in.
• Verified with targeted pytest / syntax checks.

How to Test

1. On dual AMD GPU host pick GPU index 1 with HIP in launch command. 2. Start llama.cpp serve from Cookbook. 3. Runner log shows HIP_VISIBLE_DEVICES=1 not CUDA.