Since there seemed to be some sublte differences in the data that was
maintained whether you ran goto-cc first (independently) or directly fed
the .c file into CBMC. These tests replicate the behaviour of tests
found in regressions/goto-instrument-typedef/