Retry CUDA 12 migration of hoomd-dlext

[jakirkham]

This PR was closed out instead of re-run. So hoomd-dlext wasn't actually migrated successfully. This requeues the hoomd-dlext CUDA 12 migration to generate a new migration PR.

xref: conda-forge/hoomd-dlext-feedstock#8

[jakirkham]

cc @beckermr

[beckermr]

This file won't effect anything. You'll have to remove the bit of json that stores the prs already made.

[jakirkham]

Thanks Matt! 🙏

Unfortunately am still unclear on what needs to be changed. When grepping for hoomd-dlext, am seeing the following files

$ git grep -l 'hoomd-dlext'
all_feedstocks.json
graph.json
node_attrs/a/a/b/4/e/hoomd-dlext.json
ranked_hubs_authorities.json
status/migration_json/cuda118.json
status/migration_json/cuda120.json
status/migration_json/pypy38.json
status/migration_json/python311.json
status/migration_json/python312.json

AFAICT all_feedstocks.json, graph.json, and ranked_hubs_authorities.json are not relevant

While node_attrs/a/a/b/4/e/hoomd-dlext.json could be relevant, wasn't seeing a mention of CUDA 12 or the PR (though maybe I'm missing something)

Of the migrations, only status/migration_json/cuda120.json looked relevant. Hence removing these lines

Is there something I should be looking at in one of these files that I've missed (or some other file I'm overlooking)? Or do the changes need to occur in a different repo?

[beckermr]

There should be a pr info entry for the pr. It appears this was deleted maybe during the rerun. You'll have an easier time migrating it by hand at this point. Debugging this in the bot is going to be painful.