CI: Automate syncing from the `conda-forge` channel to `nvidia`?

[leofang]

Is this feasible at all?

[leofang]

cc @cryos

[rparolin]

@cryos Is this feasible to implement in the short term?

[cryos]

Is this feasible at all?

I took a quick look around and see conda-mirror which looks like it does most of what we want. There doesn't seem to be a direct API (some quick AI chats didn't turn up anything useful). I had wanted to automate this locally at least and then translate it into a workflow.

[leofang]

IIUC this is what conda-forge uses
https://github.com/conda-forge/conda-forge-webservices/blob/8270ca82920d7585ef20c0670d3b97351b20f353/conda_forge_webservices/feedstock_outputs.py#L276-L313
For security reasons, conda-forge first pushes a package to the cf-staging channel, then copies it to cf-post-staging, finally to conda-forge:
https://conda-forge.org/docs/maintainer/infrastructure/#output-validation-and-feedstock-tokens