Provide saturating integer operations for internal use

[allight]

For analysis of tuple and array types we often need to convert and perform math on int64 values. In these cases we do not want the math to overflow. It would be useful to have a full suite of 'saturating' math ops in the math_util.h file.