Recording the expected and actual checksum in the error makes it possible to
provide that information, for example in a failed test like the ones for DRA.
Otherwise developers have to manually step through the test with a debugger to
figure out what the new checksum is.