copy all fmha headers when building library (#1497)

* copy all fmha headers when building library

* fix the rocm_install call for mha headers

library/src/tensor_operation_instance/gpu/mha/CMakeLists.txt

@@ -18,17 +18,10 @@ endif()
 
 rocm_install(DIRECTORY ${CK_TILE_SRC_FOLDER} DESTINATION ${CMAKE_INSTALL_INCLUDEDIR}/ck_tile)
 
-rocm_install(FILES
-    "${FMHA_SRC_FOLDER}/fmha_fwd.hpp"
-    "${FMHA_SRC_FOLDER}/bias.hpp"
-    "${FMHA_SRC_FOLDER}/mask.hpp"
-    DESTINATION include/ck_tile/ops
-)
-
-# header for building lib
-file(COPY ${FMHA_SRC_FOLDER}/fmha_fwd.hpp DESTINATION ${FMHA_CPP_FOLDER})
-file(COPY ${FMHA_SRC_FOLDER}/bias.hpp DESTINATION ${FMHA_CPP_FOLDER})
-file(COPY ${FMHA_SRC_FOLDER}/mask.hpp DESTINATION ${FMHA_CPP_FOLDER})
+file(GLOB MHA_HEADERS "${FMHA_SRC_FOLDER}/*.hpp")
+rocm_install(FILES ${MHA_HEADERS} DESTINATION include/ck_tile/ops)
+# headers for building lib
+file(COPY ${MHA_HEADERS} DESTINATION ${FMHA_CPP_FOLDER})
 
 # generate a list of kernels, but not actually emit files at config stage
 execute_process(