Topic: minimax m3

The ggml/llama.cpp project has released version b10288, re-enabling MiniMax M3 tests in `test-llama-archs`. This update is significant for developers and users of the llama.cpp library as it ensures compatibility and functionality across different architectures, particularly on macOS Apple Silicon.