* Previously the JdbcSource was combining the columns of any tables with the same name across different schemas into a single stream in the catalog.

* This was caught because in those tables there were columns of the same name with different types which triggered a precondition to check for this.

* The fix makes sure we group by both schema name and table name.

* Adds test to the standard jdbc tests to catch this case.

* This test does NOT run for mysql as, mysql has no concept of schemas.