Add support for DROP SCHEMA CASCADE in JDBC connectors

[ebyhr]

Description

Relates to #17649

Release notes

(x) Release notes are required, with the following suggested text:

# ClickHouse, MariaDB, MySQL, Oracle, SingleStore
* Add support for `CASCADE` option in `DROP SCHEMA` statement. ({issue}`18305`)

[ebyhr]

TestPostgreSqlConnectorTest.testSelectInformationSchemaColumns failed though the test succeeded on the 2nd attempt. Going to remove support for PostgreSQL and Redshift from this PR.

Error:  Failures: 
Error:    TestPostgreSqlConnectorTest>BaseConnectorTest.testSelectInformationSchemaColumns:2244->AbstractTestQueryFramework.assertQuery:339 Execution of 'actual' query 20230715_051130_02304_yh37n failed: SELECT table_name, column_name FROM information_schema.columns WHERE table_catalog = 'postgresql' AND table_schema = 'tpch' AND table_name LIKE '_rders'

The cause might be concurrent set column type tests.

2023-07-14T23:11:31.115-0600	INFO	pool-5-thread-2	io.trino.testng.services.ProgressLoggingListener	[TEST SUCCESS] io.trino.plugin.postgresql.TestPostgreSqlConnectorTest.testSetColumnTypeWithDefaultColumn; (took: 0.2 seconds)
2023-07-14T23:11:31.115-0600	INFO	pool-5-thread-2	io.trino.testng.services.ProgressLoggingListener	[TEST START] io.trino.plugin.postgresql.TestPostgreSqlConnectorTest.testSetColumnTypeWithNotNull
2023-07-14T23:11:31.116-0600	ERROR	SplitRunner-9-172	io.trino.execution.executor.TaskExecutor	Error processing Split 20230715_051130_02304_yh37n.0.0.0-0 io.trino.connector.informationschema.InformationSchemaSplit@7259262 (start = 526942.025557, wall = 42 ms, cpu = 0 ms, wait = 0 ms, calls = 1): JDBC_ERROR: Error listing table columns for catalog postgresql: ERROR: could not open relation with OID 18382

https://github.com/trinodb/trino/actions/runs/5559234810/jobs/10156058413?pr=18305

[hashhar]

Generally looks good.

[hashhar]

Do we want to prevent this if cascade was not specified? (and in MySQL, MariaDB and Singlestore too).

[ebyhr]

Do you mean throwing an exception when !cascade? It means users always need to specify cascade option even when the target database is empty, right?

[hashhar]

What I was thinking of would require a list operation - and even then there can be races.
I don't know a good solution. My main concern is that if someone does a DROP SCHEMA (without CASCADE) they may not expect the operation to actually CASCADE to all objects in the schema.

[ebyhr]

Added the test and manually tested without the same check in DropSchemaTask.

[hashhar]

LGTM. Squash fixups.