compile with -fno-strict-aliasing

+ with -O2, -fstrict-aliasing is enabled by default, which may cause undefined behavior (since C2011) when we dereference e.g. uint16_t[4] as a double
+ avoid undefined behavior by disabling this optimisation
+ in the future it would be probably better to use standard-supported methods for conversions (to be determined)

cmake/compiler.cmake

@@ -12,7 +12,7 @@ function(CHECK_CXX14)
   set(CMAKE_CXX_FLAGS ${CXX14_FLAG} PARENT_SCOPE)
 endfunction()
 
-set(GXX_ADDITIONAL_FLAGS "-O2 -g -ggdb -fstack-protector-strong -Wall -D_FORTIFY_SOURCE=2")	#  -fPIC is added differently, -DPIC is unknown by g++
+set(GXX_ADDITIONAL_FLAGS "-O2 -g -ggdb -fno-strict-aliasing -fstack-protector-strong -Wall -D_FORTIFY_SOURCE=2")	#  -fPIC is added differently, -DPIC is unknown by g++
 set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} ${GXX_ADDITIONAL_FLAGS}")
 
 # GCC has this options, clang doesn't