ci: fix waiting for queries in stress check

[azat]

Right now it always fails with timeout:

   Received exception from server (version 25.7.1):
   Code: 159. DB::Exception: Received from localhost:9000. DB::Exception: Timeout exceeded: elapsed 60033.770956 ms, maximum: 60000 ms. (TIMEOUT_EXCEEDED)
   (query: SELECT sleepEachRow((
           SELECT maxOrDefault(300 - elapsed) + 1
           FROM system.processes
           WHERE query NOT LIKE '%FROM system.processes%' AND elapsed < 300
       ) / 300)
       FROM numbers(300)
       FORMAT Null
       SETTINGS function_sleep_max_microseconds_per_block = 0
       )

Rewrite it to sleep only until there are some queries

CI: https://s3.amazonaws.com/clickhouse-test-reports/json.html?
PR=82364&sha=latest&name_0=PR&name_1=Stress+test+%28amd_debug%29

Changelog category (leave one):

• CI Fix or Improvement (changelog entry is not required)

[clickhouse-gh]

Workflow [PR], commit [0ac3781]

Summary: ❌

job_name	test_name	status	info	comment
Stress test (amd_tsan)		failure
	Server died	FAIL
	Hung check failed, possible deadlock found (see hung_check.log)	FAIL
	Killed by signal (in clickhouse-server.log)	FAIL
	Fatal message in clickhouse-server.log (see fatal_messages.txt)	FAIL
	Killed by signal (output files)	FAIL
	Found signal in gdb.log	FAIL

[Algunenano]

Any idea why we don't KILL all queries before waiting? It'd be faster in most cases (and when it doesn't it's a misbehaviour)

[azat]

Any idea why we don't KILL all queries before waiting? It'd be faster in most cases (and when it doesn't it's a misbehaviour)

No clue, but if it works now w/o hanged queries, then it make sense to preserve to make sure that all queries can be finished even after stressing

[azat]

Stress test (amd_tsan) — Killed by signal (in clickhouse-server.log)

• [LOGICAL_ERROR] Column {} query tree node does not have valid source node #82092