Intervals between fake HSI events are not constant

[philiprodrigues]

I notice that the timestamps of fake HSI events are not separated by a constant, but instead vary by some amount, and are even sometimes zero (ie, successive fake HSI events have the same timestamp). I think it would be better for test reproducibility if the timestamps were separated by a constant value. I'm happy to write the code to make this change if you agree.