llama.cpp Release b11277 Fixes Memory Leaks in ggml-zdnn Backend

llama.cpp released build b11277, introducing buffer resets and addressing memory leaks within the ggml-zdnn backend. The update was contributed by IBM engineer Aaron Teo along with minor code style formatting fixes. This patch release enhances memory stability and resource management for enterprise environments running LLM inference on IBM mainframes. It ensures long-running inference tasks utilizing the zDNN hardware acceleration library do not suffer from memory accumulation. The release centers around PR #29637, which implements buffer reset routines directly into ggml-zdnn. Binaries have been automatically built and published across supported platforms, including Linux, Windows, macOS, Android, and Snapdragon devices.