Revert "Image: silence compiler warnings"

This reverts commit afa0a7dac8.

src/logo/image/im6.c

@@ -1,9 +1,10 @@
 #ifdef FF_HAVE_IMAGEMAGICK6
 
 #include "image.h"
-#include "magickcore_wrapper.h"
 #include "common/library.h"
 
+#include <magick/MagickCore.h>
+
 static FF_LIBRARY_SYMBOL(ResizeImage)
 
 static void* logoResize(const void* image, size_t width, size_t height, void* exceptionInfo)

src/logo/image/im7.c

@@ -1,9 +1,10 @@
 #ifdef FF_HAVE_IMAGEMAGICK7
 
 #include "image.h"
-#include "magickcore_wrapper.h"
 #include "common/library.h"
 
+#include <MagickCore/MagickCore.h>
+
 static FF_LIBRARY_SYMBOL(ResizeImage)
 
 static void* logoResize(const void* image, size_t width, size_t height, void* exceptionInfo)

src/logo/image/image.c

@@ -265,7 +265,11 @@ static bool compressBlob(void** blob, size_t* length)
 #endif // FF_HAVE_ZLIB
 
 //We use only the defines from here, that are exactly the same in both versions
-#include "magickcore_wrapper.h"
+#ifdef FF_HAVE_IMAGEMAGICK7
+    #include <MagickCore/MagickCore.h>
+#else
+    #include <magick/MagickCore.h>
+#endif
 
 typedef struct ImageData
 {

src/logo/image/magickcore_wrapper.h

@@ -1,13 +0,0 @@
-#pragma once
-
-// ImageMagick-7/MagickCore/pixel-accessor.h:518:46: warning: implicit conversion from 'QuantumAny' (aka 'unsigned long long') to 'double' may lose precision
-#pragma GCC diagnostic push
-#pragma GCC diagnostic ignored "-Wimplicit-int-float-conversion"
-
-#ifdef FF_HAVE_IMAGEMAGICK7
-    #include <MagickCore/MagickCore.h>
-#elif defined(FF_HAVE_IMAGEMAGICK6)
-    #include <magick/MagickCore.h>
-#endif
-
-#pragma GCC diagnostic pop