Possible Improvement of constructDuration.py Script

[VincenzoFerme]

The constructDuration.py is the slowest one. Strangely it is slow also at experiment level, although I expect it to be really fast given that it simply takes already computed metrics from the trial metrics.

For the trial level evaluate the following improvements:

• Reuse the filteredRDD instead of basically computing it again, and also reuse the data variable where you basically have all the information you compute again for the for