LibGfx: Fix writing PNG headers on x86_64

m_data.size() returns a size_t which is a 64-bit type on x86_64. This
resulted in us incorrectly using zero in the PNG header.

Userland/Libraries/LibGfx/PNGWriter.cpp

@@ -79,7 +79,7 @@ void PNGChunk::store_type()
 
 void PNGChunk::store_data_length()
 {
-    auto data_length = BigEndian(m_data.size() - sizeof(data_length_type) - m_type.length());
+    auto data_length = BigEndian<u32>(m_data.size() - sizeof(data_length_type) - m_type.length());
     __builtin_memcpy(m_data.offset_pointer(0), &data_length, sizeof(u32));
 }