Managing configuration and script updates

[Cerfoglg]

Given that the analyser and data transformer scripts, as well as the configuration for the STS can change, we need a way to stop the metric computation. However, if the computation of the metrics for a given experiments have been started, meaning that 1 or more trials of that experiments have had their metrics computed, we have to finish the computation for that experiment before stopping. We also need to save the state of the computations after all is completed, which include kafka messages not belonging to the experiment we are finishing up.

Note: this is a simplification, because not all the changes to the configuration and/or scripts impact the consistency of computed metrics. A more fine-grained analysis will be part of future development.