2 sides of that coin:
- drop the R cran script
- parse the listing page instead (as in simple_lister, check lister cgit's way of doing it) 
- for each package found there, send the origin url  to the loader (as `recurring` task)
- make the tasks outputed by the lister as `recurring` (currently `oneshot`)
- Adapt uid field to be the origin_url's value
- truncate cran_repo table
- trigger back a full listing
- Improve the loader so it scrapes that origin url  page.
- It then determines itself what the artifact urls it needs to ingest
- In the  page, there is an archive link `Old source` which lists the previous artifact version.
The good news is this that be done independently and in any order (this task can then be split in 2 subtasks).
This can be subject to discussion with the cran community to ask for a better api endpoint (if it's not too much hassle for them to adapt and provide ;)