CEP: document/standardize repodata hotfixing process; extend to run_exports

[msarahan]

Checklist

• I added a descriptive title
• I searched open requests and couldn't find a duplicate

What is the idea?

Allow fixing mistakes in run_exports metadata

Why is this needed?

In July 31 2024 meeting, I expressed interest in adding run_exports hotfixing capabilities. @jezdez pointed out that the existing hotfixing capability for repodata.json is not well documented, and also differs between Anaconda and Conda-forge.

@jaimergp proposed conda/conda-index#102 and implemented this in CEP 12 and conda/conda-index#110. As Jaime mentioned, this opens the door to hotfixing run_exports metadata when conda-index gets run.

At time of writing, I'm not sure that the actual process of generating and applying patches has been done. In discussing this, @wolfv said that he'd really like to have run_exports information be co-located with the repodata.json file. This costs some file size space in repodata.json, so it has been undesirable. However, with sharding by package name, the size impact will be relatively small, and the simplicity of having the run_exports metadata in the main repodata.json for each shard is worthwhile.

What should happen?

• Document the current process that is used when hotfixing repodata.json for conda-forge
• Propose a process for hotfixing run_exports
• Define a migration path for combining run_exports metadata into repodata.json (may require sharding to be implemented first)

Additional Context

No response