ggerganov/llama.cpp Released Build b11565 with SYCL Fix

llama.cpp release b11565 introduces a specific bugfix for SYCL to properly guard DMMV reads at row tails. Additionally, it provides standard cross-platform build artifacts for various operating systems and hardware configurations. This update improves stability and execution reliability for users running large language models on hardware accelerators via the SYCL backend. Such incremental maintenance releases ensure the broad ecosystem of local AI tools relying on llama.cpp remains robust. The update focuses entirely on a minor code adjustment under pull request #29338, preventing out-of-bounds memory access issues during row-tail processing in DMMV operations.