llama.cpp b9979 fixes silent prompt truncation on embedded NUL

The llama.cpp project released version b9979, which includes a fix for silent prompt truncation when messages contain embedded NUL bytes. Previously, the `mtmd_input_text` function carried prompts as bare pointers without length information, causing content after a NUL byte to be dropped silently.

• The fix adds an explicit `text_len` parameter to `mtmd_input_text`, ensuring the full prompt is passed through correctly.
• This change aligns with how `llama_tokenize` and the text-only path handle input lengths.
• Binaries are available for macOS (Apple Silicon and Intel), iOS, Linux (CPU, Vulkan, ROCm, OpenVINO, SYCL), Android, Windows (CPU, OpenCL, CUDA 12/13, Vulkan, OpenVINO, SYCL, HIP), and openEuler.

This update prevents data loss in prompts containing NUL characters, ensuring complete message processing across all supported platforms.