Pretty sure you can use the Spark driver for SQL Server on Shared UC clusters. 

More info here: https://learn.microsoft.com/en-us/sql/connect/spark/connector?view=sql-server-ver16

I do agree that you should rather use lakehouse federation, although you won't be able to do upserts in SQL this way. It can only do read operations. 

Moving from Non-Isolation Shared to Shared usually comes with a lot of (legacy) code refactoring.