llama.cpp b9956 fixes CLI crash on invalid server URL

The latest release of llama.cpp, tag b9956, addresses a critical crash bug in the command-line interface. When the server base URL was incorrectly specified (e.g., a typo or unreachable endpoint), the program would crash instead of gracefully handling the error. The fix, merged via PR #25497, catches JSON errors and standard exceptions, allowing the CLI to exit cleanly with a user-friendly message. This is especially important for automated scripts and deployment pipelines where a misconfigured URL could otherwise cause silent failures.

Beyond the crash fix, b9956 continues llama.cpp's tradition of broad platform support. Builds are available for macOS (Apple Silicon with optional KleidiAI acceleration, Intel x64, iOS XCFramework), Linux (x64, arm64, s390x, plus Vulkan, ROCm 7.2, OpenVINO, SYCL FP32/FP16), Windows (x64 CPU, arm64 CPU, arm64 OpenCL Adreno, CUDA 12/13, Vulkan, OpenVINO, SYCL, HIP), and Android arm64. This release is signed with GitHub's verified GPG key (B5690EEEBB952194), ensuring integrity. For developers and enthusiasts running local LLMs, updating to b9956 eliminates one more potential point of failure.