Parallelize optimization and code generation for link time optimization.

This parallelizes the `-flto` feature resulting in parallel optimization
and code generation for link time optimization based on the number of
logical processors available.

Note: this has the side-effect of using more memory during linking.

References:

- https://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html (see: -flto)
- http://www.cmake.org/cmake/help/v3.0/module/ProcessorCount.htmMC

CMakeLists.txt

@@ -124,14 +124,16 @@ endif()
 if(CMAKE_BUILD_TYPE MATCHES Release)
   message(STATUS "Configuring OSRM in release mode")
   # Check if LTO is available
+  include(ProcessorCount)
+  ProcessorCount(NPROC)
   set(LTO_FLAGS "")
-  check_cxx_compiler_flag("-flto" LTO_AVAILABLE)
+  check_cxx_compiler_flag("-flto=${NPROC}" LTO_AVAILABLE)
   if(LTO_AVAILABLE)
-    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -flto")
+    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -flto=${NPROC}")
     set(CHECK_LTO_SRC "int main(){return 0;}")
     check_cxx_source_compiles("${CHECK_LTO_SRC}" LTO_WORKS)
     if(LTO_WORKS)
-      message(STATUS "LTO working")
+    message(STATUS "LTO working")
     else()
       message(STATUS "LTO broken")
       set(CMAKE_CXX_FLAGS "${OLD_CXX_FLAGS}")