There were a bunch of issues in the tracking of logical and physical successors and predecessors in the Intel compiler. This fixes a few issues, but I suspect there are still more. It's enough to get !25132 (closed) working anyway.
I also have a commit that adds some validation for
ENDIF instructions. It fails assertions all over the place. I haven't determined if the CFG is wrong or the validation tests are wrong. I don't have any more time to spend on it right now, so I'm setting any further validation to the side for now.