Conteneurisation of output -- Thoughtput calibration

Summary

Thoughput calibration needs to use containers to store and share its results.

What is the expected correct behavior?