Crikey, this raises more questions than answers...
So...
"IBM's claims that it just needed a little light tweaking. IBM hotly denies that it was unsuitable."
Having worked on systems for both US and UK insurers, I very much doubt it was suitable. There are quite significant differences in the ways that insurance books are managed between the two.
"The system is in fact performing as expected"
In what way? Functionally or non-functionally? And according to who? IBMs assumptions? Or as per the end clients use cases, FRs and NFRs?
And finally re the data migration, surely IBM would have known about this 3rd party risk or dependency at the beginning of the contract so I'm struggling to understand why they didn't manage it as such : using this as a reason to walk away seems a little weird.