build(cmake): disable building of upstream unified binary

Set `LLAMA_BUILD_APP` to `OFF` to prevent the compilation of the new
unified `llama` binary introduced in upstream llama.cpp.

Since the Python package only requires the underlying shared libraries
and specific targets, explicitly disabling the standalone application
reduces build times and prevents unnecessary executable artifacts from
being compiled.

Signed-off-by: JamePeng <jame_peng@sina.com>

Author: JamePeng

CMakeLists.txt

@@ -222,6 +222,9 @@ if (LLAMA_BUILD)
     # Disable building of server
     set(LLAMA_BUILD_SERVER OFF CACHE BOOL "llama: build server example" FORCE)
 
+    # Disable building of unified binary
+    set(LLAMA_BUILD_APP OFF CACHE BOOL "llama: build the unified binary" FORCE)
+
     # Disable build the embedded Web UI for server
     set(LLAMA_BUILD_UI OFF CACHE BOOL "llama: build the embedded Web UI for server" FORCE)
     set(LLAMA_USE_PREBUILT_UI OFF CACHE BOOL "llama: use prebuilt UI from HF Bucket when available (requires LLAMA_BUILD_UI=ON)" FORCE)