Define the job and what it owes. Point at the raw capture in Drive — the files are tens of gigabytes, so we link them rather than move them through a browser. Everything here can be changed later.
What this job owes when it is finished. This is what the job page measures itself against, so a missing file reads as outstanding rather than as an absence nobody noticed.
Creating the job records it and its deliverables. Processing is started by the rig — this does not kick off a run on its own yet.