Separate database for preaggregations

I am using CubeJS against read-only replica of the MS SQL database (I imagine it is a very common scenario).
Regardless of schema, the database is read-only, so CubeJS can not create temporary preaggreation tables.

I need a way to specify a database name (on the same server) for preaggregations. I tried providing preaggregation schema with '.' in it (i.e. 'cubejsdb.preagg_schema') but got an error about table name having an extra '.' or something like that.

I was able to work around this issue as follows:
1. Point CUBEJS_DB_NAME to an empty R/W database that will receive preaggregations
2. In schema files add a dbname prefix to all tables: Instead of 'select .. from my_table1 join mytable2 ' I have 'select ... from actual_db..my_table1 join actual_db..my_table2' . (where actual_db is my R/O replica)

Since the databases are on the same server, 'insert from select' statements that cubejs uses work fine.
This solution is messy since I have to deploy these template files to various environments that have possibly different database names.