feat: add constraint to deny drop sources referenced by other CREATE_AS sources

[spena]

Description

This is partial implementation of feat: terminate persistent query on DROP command

The first commit of this PR was already reviewed in the above PR (#6143). This PR only adds extra work to:

• Return DROP source constraints from DESCRIBE EXTENDED
• Drop sources in order for Rest QTT cleanup methods

The change for this PR adds a constraint to a source to not delete it if it is referenced by another source. For instance:

ksql> create stream t1(id int) with (kafka_topic='t1');
ksql> create stream s1 as select * from t1;
ksql> terminate CSAS_S1_9;

ksql> drop stream t1;
Cannot drop T1.
The following streams and/or tables read from this source: [S1].
You need to drop them before dropping T1.

ksql> describe extended t1;
...
Sources that have a DROP constraint on this source
--------------------------------------------------
S1
...

Even if queries are terminated, the stream or table cannot be deleted until all sources that reference to it are deleted. This is a common approach in other DBs, which could keep constraints and references on tables thus avoiding deleting them leaving tables with an invalid link.

This feature was asked by #6143. The main reason of splitting the PR is for easy review the rest of the commits.

Commit #1 is already reviewed and approved. If you want you can review again, but you can opt-out, and review only Commit #2 and Commit #3, which add the list to the DESCRIBE EXTENDED and fixes the RQTT tests. The rest of the commits are just fixing a couple of tests.

Testing done

Describe the testing strategy. Unit and integration tests are expected for any behavior changes.

Reviewer checklist

• Ensure docs are updated if necessary. (eg. if a user visible feature is being added or changed).
• Ensure relevant issues are linked (description should include text like "Fixes #")

[agavra]

LGTM!

[agavra]

what happens if we get old response with new CLI and vice versa?

[spena]

From an old CLI to the new Server works.

From the new CLI to the old Server doesn't work. But, is this a valid support? And, how do others maintain compatibility with an old Server? I was looking at the code for other properties and I don't see how this is supported. Any idea?

[agavra]

From an old CLI to the new Server works.

This one is more important. The other is a nice-to-have. It's not expected to work, but it would be nice if it behaved nicely (either ignoring it or telling you there's an issue). You don't need to implement that for this pr

[spena]

Btw, I added a new test in RecoverTest to validate the scenario caused by an upgrade, where the DROP stream could be in an order where it has DROP constraints. There was nothing to fix. The compactor did the thing to remove the query, thus removing the DROP constraint. I'll make sure to re-validate this before merging the DROP+TERMINATE command.