In HELP-97969 and HELP-98033 it seems that segments with no phylogs are never marked fully materialized, which leads to materialization outages.

Mongo has some of the information needed to determine if a segment is empty and assume that it is materialized if so (by comparing seal LSNs between segments).

This only works if we have boundaries for the segment before the empty one, which we won't have for deleted segments or the 0th segment (which HELP-98033 needs) so its possibly only LS could have this information. We may need to advocate for adding it to the seal record.