Discuss approaches for dealing with missing version data

E.g., due to a broken scraper for some periods of time, or too infrequent of updates. Interacts with planning for datetime versions. E.g., in another issues, we have discussed an approach where an archive also keeps track of which versions were recorded. We could require a recorded version within some amount of time before each requested ref_time_values for a slide or as_of for an as_of operation, and it would attack both issues.

The issue names no files, tests, or entry points. Start by reviewing the existing archive and version-handling behavior, along with the planning for datetime versions described here. Done would require agreeing on a concrete approach for detecting or handling missing recorded versions and documenting the resulting requirements.

Written by the indexing model from the issue text.