fix: Calculate multiples using two ways and take the average to adjust for floating point representation issues

[waggonerjake]

Closes #2962

[Marsup]

Hi,

While this works, I'm wondering which method you based this patch on, do you have any academic reference that would prove this always works?

Also, I'm a bit worried about overflows, as adding big numbers can result in this scenario.

[Marsup]

Closing for lack of answer.