b9102: [SYCL] Add OP im2col_3d (#22903)

Release b9102 adds support for the im2col_3d operator to the SYCL backend in llama.cpp, expanding GPU-accelerated tensor operations for Intel GPU inference. The im2col_3d operation converts 3D input volumes to column matrices, a fundamental computation in convolutional neural networks that was previously missing from SYCL kernel implementations. This enhancement improves compatibility with models that use 3D convolution operations on Intel GPUs.

Source: llama.cpp Releases | 2026-05-11