Definitely a funky embedded font situation. Sure would like to get a hold of a legal, non-DRMed KF8 file that exhibits this behavior to see exactly what's going on.
To my knowledge... v0.54 should preserve the entire raw resource record as a *.failed file in the unpacked data. That was done in the hopes that someday, someone would be able to take a look at it and see what might be going on. I'd settle for the file-size of the .failed file, for starters, if anyone comes across the situation.
