pub fn try_shared_gpu() -> Option<&'static SharedGpuContext>Expand description
Process-wide wgpu device + queue, or None when no usable GPU adapter exists
(headless / integrated-only / GPU-less machine, or the tokio runtime can’t start).
Callers that can degrade to CPU MUST use this and fall back on None, rather than
shared_gpu() which aborts the process. The None result is cached, so a GPU-less
machine probes once. (Init is lazy and reused by QTensorEngine + render.)