Summary

Studio's GPU Monitor currently surfaces VRAM / power / temperature for only GPU 0, even on systems with multiple NVIDIA GPUs.

Problem

On multi-GPU workstations and servers, that makes the monitor misleading:

• active work may be running on a different GPU
• users cannot see total GPU count from the Studio UI
• long GPU names and metric strings can overflow the current compact card layout

Proposal

Update the Studio GPU monitor so it:

• returns telemetry for every visible GPU from the backend
• exposes the GPU count in the frontend
• keeps the existing top-level GPU 0 fields for compatibility
• renders one responsive card per GPU in the progress panel
• avoids text overflow for long GPU names and metric values

Notes

I already have a branch prepared with the backend payload change, tests, and a responsive frontend layout for multi-GPU systems.