It's not just validation. A lot of things compilers have to do becomes dramatically simpler with structured control flow and simpler means faster. When you operate on an unstructured CFG you first have to analyze the entire graph to find the dominators, loops etc and then you have to propagate the dataflow information in that order. All these complications go away.
There's been a few other works that realized this, among them the CORTL IR, suggested by Carl McConnell in his Thesis proposal "Tree Based Optimizations" and the IR used in the Oberon compiler by Michael Franz.