Missing public header for ggml-backend.h and ggml-alloc.h

[65a]

The public header defined in CMakeLists.txt for llama.h (used for library level C FFI integrations) depends on the header ggml-backend.h now, but as it is not a defined public header in CMakeLists.h, it cannot be found in the CMake output. It seems like this is an omission, and the CMakeLists.txt should include this file as a public header.

[65a]

Can add ggml-alloc.h to the list as well.

Builds and links correctly with something like this:

diff --git a/CMakeLists.txt b/CMakeLists.txt
index 7bd64096..088245fe 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -846,7 +846,7 @@ install(FILES ${CMAKE_CURRENT_BINARY_DIR}/LlamaConfig.cmake
               ${CMAKE_CURRENT_BINARY_DIR}/LlamaConfigVersion.cmake
         DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/Llama)

-set(GGML_PUBLIC_HEADERS "ggml.h"
+set(GGML_PUBLIC_HEADERS "ggml.h" "ggml-backend.h" "ggml-alloc.h"
         "${GGML_HEADERS_CUDA}" "${GGML_HEADERS_OPENCL}"
         "${GGML_HEADERS_METAL}" "${GGML_HEADERS_MPI}" "${GGML_HEADERS_EXTRA}")

[65a]

Thanks!