Fix handling error when failing to drop the database completely on the cluster on the secondary node.

[tuanpach]

Changelog category (leave one):

• Bug Fix (user-visible misbehavior in an official stable release)

Changelog entry (a user-readable short description of the changes that goes into CHANGELOG.md):

Fix handling error when failing to drop the database completely on the cluster on the secondary node.

There are 2 problems in https://github.com/ClickHouse/support-escalation/issues/5792:

1. When dropping a DatabaseReplicated in (DatabaseReplicated::drop), the process fails when trying to remove the Keeper path.
   In DatabaseCatalog::detachDatabase, we don't handle the exception, which leads to:

• The database metadata file is still on disk. It will fail if users create a new database with the same name.
• Database is not re-attached, so "DROP DATABASE IF EXISTS" will succeed but have no effect.

2. When dropping the database with "ON CLUSTER", DDLWorker doesn't handle retry properly. In fact, it just writes down the status on Keeper. We can try to do the task if the task fails and is retriable.

The change in this PR:

• Re-attach the database if it failed to drop the database completely
• In DDLWorker, when the task is not finished, throw an ErrorCodes::UNFINISHED exception if the task is retriable to retry the task.

Closes https://github.com/ClickHouse/support-escalation/issues/5792

Documentation entry for user-facing changes

• Documentation is written (mandatory for new features)

[clickhouse-gh]

Workflow [PR], commit [8c2cc4c]

Summary: ❌

job_name	test_name	status	info	comment
Bugfix validation (integration tests)		error