Document when the Execution Time can be zero for the Throughput metric at trial level

[VincenzoFerme]

Document when can the following happen (ex = 0): https://github.com/benchflow/analysers/pull/90/files#diff-719a3c15574b29171190bd83ee789a1aR33

[Cerfoglg]

@VincenzoFerme

It shouldn't happen, but in the off chance that an execution time is lower than the precision level of the reported execution times it's good to have that. But it might be overly zealous in that regard.