Users that upgrade from earlier versions of the server may contain mixed-schema data in their time-series buckets. As long as the catalog flag that indicates whether a time-series bucket may contain mixed-schema data is set to true, we need to validate all the buckets when trying to index time-series measurements.

There are two possible outcomes:

1. There are no buckets containing mixed-schema data. We set the catalog flag to false and subsequent index builds do not have to perform any validation on the buckets.
2. There is at least one bucket containing mixed-schema data. The index build will fail in this case. We'll also provide a list of options for migrating the time-series buckets to a format that does not contain problematic mixed-schema data.