"Scripts" in this case is just the script function in bin/gargantext-client/Script.hs, which currently contains a rather simple initial example of hitting the API, and tracking some GHC RTS allocation metric while doing a bunch of queries to the garg server. We can decide what a more complete/useful/helpful example would be in this MR or in a subsequent one, I don't mind.

All the infrastructure is there to use the entirety of the backend's API in those scripts, from tweaking node related things to submitting and polling jobs. If the steps were to be timed, we could also track duration of each step and be able to observe improvements through allocation metrics, GC cycles/time, execution time. We can build some higher level blocks for managing jobs or generating some interesting initial state if facing a fresh/local garg instance. You get the idea, this is pretty flexible and can be used to support various "experiments", as we've described before.

I tried to be generous with the comments.