Either that or something more evolved as discussed in the related task.
I'd argue that i'd very much like to close the bitbucket origin ingestion which has been
ongoing for months now. So that naive approach suits me well enough right now.
We can always evolve this a bit more later. There is a dedicated opened task btw to
actually discuss on how to improve the coverage setup and avoid those kind of static
dicts (i don't find back the reference).
For this to evolve here, that'd require at least an evolution in the scheduler 'listers'
table to have a 'discontinued' flag (or something with a better name) that we could lift
to actually query on {lister name, instance name, visit_type}. That'd also require we
reference actual origins properly in listed_origins. And that could be done rather
easily for the bitbucket mercurial origins, but not necessarily for the other ones (in
the modified dict from the diff).
Related to T3658