I don't think that casting a range of bits as some other arbitrary type "is a bug nobody sees coming".
C++ compilers also warn you that this is likely an issue and will fail to compile if configured to do so. But it will let you do it if you really want to.
That's why I love C++
The problem is that it's undefined behavior. Quake fast inverse square root only works before the types just happen to look that way. Because the floats just happens to have that bit arrangement. It could look very different on other machines! Nevermind that it's essentially always exactly the same on most architectures. So yeah. Undefined behavior is there to keep your code usable even if our assumptions about types and memory change completely one day.