I'm curious, what would be a good reason why treating floating point numbers as byte sequences should be any harder than what is required to make it obvious (provided their binary format is well defined)?
There are footguns in making that representation easy to access, e.g. if you try to hash the byte sequence to use floats as hash table keys then it will almost work but you'll get a very subtle error because 0 and -0 will hash differently. And frankly most of the things you'd do with the byte sequence are things that there are more semantically correct ways to do. There should be a way to access that representation but it shouldn't be something you'd stumble into doing accidentally, IMO.
