Rich future use long long int for truncation

Apologies @cattanem for this late MR, but just now whilst working on LHCb!603 (merged) I noticed a small issue with double truncation, that is resolved by using long long int instead of just int. In practise this makes no difference in the cherenkov angle truncation, due to the values used there as opposed to the LHCbMath test, but just to be safe this MR switches it to use the 64 bit int in any case.