fix: use correct gpu id in pod create help examples

[TimPietruskyRunPod]

Summary

• GPU ID fix: help examples used "NVIDIA RTX 4090" which is not a valid GPU ID and causes pod creation to fail. Corrected to "NVIDIA GeForce RTX 4090" (from runpodctl gpu list)
• Docker image fix: replaced outdated runpod/pytorch:2.1.0-py3.10-cuda11.8.0-devel-ubuntu22.04 with runpod/pytorch:1.0.3-cu1281-torch291-ubuntu2404
• Regenerated docs: docs/ markdown files regenerated (also picks up missing --network-volume-id flag from PR feat: add --network-volume-id flag to pod/serverless create #259)

Also needs fixing (separate repos)

• Skill (runpod/skills/runpodctl/SKILL.md): same GPU ID + image + completion instructions — fix ready locally, needs its own PR
• Docs PR (docs: Update runpodctl documentation for v2.0 docs#563): uses the same wrong GPU ID and image across multiple pages

Verified

# correct id — works
runpodctl pod create --template-id runpod-torch-v21 --gpu-id "NVIDIA GeForce RTX 4090"
# → created successfully, then deleted

# wrong id — fails
runpodctl pod create --template-id runpod-torch-v21 --gpu-id "NVIDIA RTX 4090"
# → "There are no longer any instances available"

Test plan

• Built binary, confirmed help text shows correct GPU ID and image
• E2E tested pod create with correct GPU ID — pod created and deleted
• Confirmed wrong GPU ID fails as expected

[promptless]

📝 Documentation updates detected!

New suggestion: Fix incorrect GPU ID in Hub publishing guide
Updated existing suggestion: Update runpodctl documentation for v2.0

---

Tip: Configure how Promptless handles changelogs in Agent Settings 📋