I'm not on the core conda-forge team, but I have some questions/issues.
1) As I understand it, we really don't want to remove packages from conda-forge. So you could certainly stop building new versions, but the old versions should remain. I can't tell from just the error message, but the bioconda build should probably be checking not jsut for the package name, but for the version.
2) I'm curious why you want to keep bioconda separate from conda-forge. Conda Forge was originally started to help bring multiple communities that were developing domain-specific packages together to share resources and avoid duplication -- surely the biomedical community has a need for many of the same packages as other communities.
But in any case, I can also see that there may be some (many) packages that are biomedical specific, so can see the need for the project to support that community. But:
I see from the docs that you recommend:
conda config --add channels defaults
conda config --add channels bioconda
conda config --add channels conda-forge
conda config --set channel_priority strict
which would put conda-forge at a higher priority than bioconda -- I think that's the source of your issue here -- if a package is in conda-forge, and you decide to add it to biconda, your users will still get the conda-forge version.
But if the idea is that folks would get general purpose packages from conda-forge, and the biomedical-specific ones from bioconda, then I'd suggest you put bioconda first in the priority order. That way, it leaves control of which packages come from bioconda under bioconda's control.
In fact, with the current setup, the problem package could be removed from conda-forge, but who knows what other duplicate package might get added to conda-forge later on? -- which would then break bioconda again.
Just my $0.02
-CHB