Large sandbox images (e.g. 852MB base image) can take ~3 minutes to
pull, exceeding the previous 120s timeout. Bump to 300s across all
surfaces: CLI watch loop, server orphan grace period, TUI ready poll,
Python SDK wait_ready default, and E2E test harness.
* feat(sandbox): add gpu sandbox scheduling support
Allow sandbox creation to request GPU resources explicitly or infer them from GPU image names. This wires GPU intent through bootstrap, validates gateway support, and adds dedicated GPU E2E coverage for follow-up cluster testing.