main
Benchmarks: raw-Vulkan-vs-Vulkore saxpy + GPU-vs-CPU bench harness
Batch API: many dispatches, one submit + Fence drain-wait guard
Mali/UMA portability: fix non-coherent transfers + expand test suite 52->82 (#1)
GoogleTest suite for Context/Buffer/Sync; multi-driver matrix via VULKORE_DEVICE (llvmpipe/RENOIR/NVIDIA)