Add Apple Vision Pro Docker teleoperation workflow

[benjiebob]

Summary

• add a reproducible Isaac Sim 6.0.1 / Isaac Lab Docker workflow for Apple Vision Pro CloudXR teleoperation
• add a headless launcher, persistent CloudXR EULA state, required runtime libraries, stable NVIDIA cgroup access, and a complete Linux workstation + Xcode deployment runbook
• add an optional synchronized RTSP spectator stream and a headset-free Isaac Lab desktop preview
• add the wheel + curated Snap Circuits workbench with GR1T2, Sharpa Wave, and ProHand task options; private/generated assets remain ignored and are prepared locally by pinned setup scripts

Validation

• pre-commit run --all-files
• pytest docker/test/test_dockerfile_nonroot.py -q — 24 passed
• shell syntax checks for all new launch/setup scripts
• Python compilation checks for all new and modified Python sources
• live validation on robolab1 with an NVIDIA RTX PRO 6000: AVP CloudXR connection, GR1T2 hand teleoperation, shared wheel/circuit scene, and moving 1280x720 H.264 RTSP spectator output

Notes

• GR1T2 is the default and is the end-to-end validated hand backend.
• Sharpa Wave and ProHand are included as experimental selectable backends; they still require headset-specific tracking/calibration validation.
• ProHand model geometry is fetched locally for evaluation/simulation under its upstream mesh license and is not redistributed by this PR.
• The current Snap Circuits workbench supports grasp-and-arrange interaction; it does not implement board latching or electrical behavior.