Fix possible crash in aggregate functions after MEMORY_LIMIT_EXCEEDED

[azat]

Changelog category (leave one):

• Critical Bug Fix (crash, data loss, RBAC) or LOGICAL_ERROR

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

Fix possible crash in aggregate functions after MEMORY_LIMIT_EXCEEDED

The problem is when ColumnAggregateFunction is reused after exception, this is the case of async_insert=1, where we are trying to load each INSERT batch, and if it fails, continue with other batches.

The problem was that once ColumnAggregateFunction::ensureOwnership() throws, it leaves the column in a broken state, since all aggregation states up to rollback_pos will be broken.

The fix is simple - just copy them from source column again.

And add a unit test.

[clickhouse-gh]

Workflow [PR], commit [1cd46f9]

Summary: ❌

job_name	test_name	status	info	comment
Stateless tests (amd_binary, old analyzer, s3 storage, DatabaseReplicated, sequential)		failure
	01079_parallel_alter_detach_table_zookeeper	FAIL	cidb, issue
Stateless tests (amd_msan, parallel)		failure
	02950_parallel_replicas_used_count	FAIL	cidb, issue
	01548_query_log_query_execution_ms	FAIL	cidb, issue
BuzzHouse (amd_debug)		failure
	Logical error: 'Inconsistent AST formatting: the query: (STID: 1941-1bfa)	FAIL	cidb, issue
BuzzHouse (amd_msan)		failure
	Server unresponsive: memory limit exceeded	FAIL	cidb, issue
AST fuzzer (amd_ubsan)		error

[novikd]

LGTM