These are metadata layers on top of file formats, expecting to process many files (where schema may change) in typically object storage, with volumes in excess

These are metadata layers on top of file formats, expecting to process many files (where schema may change) in typically object storage, with volumes in excess of a petabyte. There is nothing preventing you from using sqlite with iceberg, for example - except implementing it (may involve implementation in your execution engine too). It already supports Parquet, Orc, and Avro (which is more row oriented than column oriented) Iceberg and Sqlite might be interesting if you wanted to colocate two tables in the same file, for example. A smart enough data access layer with an appropriate execution engine could possibly see the data for both.