VideoBackends/OGL: Don't call glMemoryBarrier without ARB_shader_image_load_store

glMemoryBarrier comes from ARB_shader_image_load_store, so on a desktop context
below GL 4.2 the pointer is null. bSupportsBBox is set from
bSupportsFragmentStoresAndAtomics, which such a driver can still have, so saving
a state reads the bounding box and calls a null pointer.

UsePersistentStagingBuffers() in OGLTexture.cpp already guards the same call with
bSupportsImageLoadStore.

Source/Core/VideoBackends/OGL/OGLBoundingBox.cpp

@@ -5,6 +5,7 @@
 
 #include <cstring>
 
+#include "VideoBackends/OGL/OGLConfig.h"
 #include "VideoBackends/OGL/OGLGfx.h"
 #include "VideoCommon/DriverDetails.h"
 
@@ -37,7 +38,7 @@ std::vector<BBoxType> OGLBoundingBox::Read(u32 index, u32 length)
   // on nVidia drivers. This is more noticeable at higher internal resolutions.
   // Using glGetBufferSubData instead does not seem to exhibit this slowdown.
   if (!DriverDetails::HasBug(DriverDetails::BUG_SLOW_GETBUFFERSUBDATA) &&
-      !static_cast<OGLGfx*>(g_gfx.get())->IsGLES())
+      !static_cast<OGLGfx*>(g_gfx.get())->IsGLES() && g_ogl_config.bSupportsImageLoadStore)
   {
     // We also need to ensure the the CPU does not receive stale values which have been updated by
     // the GPU. Apparently the buffer here is not coherent on NVIDIA drivers. Not sure if this is a