Maybe I'm misunderstanding the feature that you would like to have, but I think Citus columnar supports what you want. Citus columnar storage has projection pushdown, so it only fetches the columns that you requested in your query.
To quote from the blogpost[1]:
What are the Benefits of Columnar?
- Compression reduces storage requirements - Compression reduces the IO needed to scan the table - Projection Pushdown means that queries can skip over the columns that they don’t need, further reducing IO - Chunk Group Filtering allows queries to skip over Chunk Groups of data if the metadata indicates that none of the data in the chunk group will match the predicate. In other words, for certain kinds of queries and data sets, it can skip past a lot of the data quickly, without even decompressing it!
Disclaimer: I work on the Citus team at Microsoft, but have not worked on the columnar storage.
[1]: https://www.citusdata.com/blog/2021/03/06/citus-10-columnar-...