BitVM is huge if true, but the abstract seems to indicate that they've solved the halting problem? In other words, that it's trivially easy to prove a script is invalid? Seems fishy to me.