Right now when query SHOW TABLES is executed in DataLakeCatalog database we do two kind of requests:

1. List all the tables, for example this one for IcebergRest: https://github.com/apache/iceberg/blob/main/open-api/rest-catalog-open-api.yaml#L516.
2. For each table from p.1 get metadata about each table, like: https://github.com/apache/iceberg/blob/main/open-api/rest-catalog-open-api.yaml#L933.

We do this because of IDatabaseTablesIterator API: https://github.com/ClickHouse/ClickHouse/blob/master/src/Databases/IDatabase.h#L51. It should be able to produce a IStorage object. However for queries like SHOW TABLES and even for most queries for system.tables it's not needed. We already have "lightweight" method for iteration over entities in databases: https://github.com/ClickHouse/ClickHouse/blob/master/src/Databases/IDatabase.h#L273-L276

However the only difference for "lightweight" iterator are some flags for StorageObjectStorage object. We still do a request to catalog for each table:

The proposal is to introduce simple structure like "StorageDescription" which will return only some parts of IStorage metadata. At least table name which will be sufficient for "SHOW TABLES" query. Also we will need to introduce new Iterator for this structure and adapt places in code where we already use "lightweight" iterator. Either use information from new really lightweight iterator, or switch to ordinary "getTablesIterator".