Improvise the column pruning pipeline

- Redefine the table schema processing into vector embeddings
- Get a fixed no of relevant columns for the query retrieval part
- Make sure that multiple user sessions can be handled
- Rethink on the type of database that is efficient 