QoQ with HSQLB creates a view to determine which source columns are used for a query.

This is an optimization for working with larger source queries to avoid loading unused columns into the temporary database used for executing the QoQ.

This involves creating a view and reading the view metadata, which adds some overhead.

This can be cached, using the actual sql and create table sql as a key (column types in lucee queries can change as data is added https://luceeserver.atlassian.net/issues/LDEV-4564 )

Goal is to address the performance regression as seen on testlab