fix: the average time for clickbench query compute should use new vec to make it compute for each query

[zhuqi-lucas]

Which issue does this PR close?

• Closes #15471

Rationale for this change

The average time for clickbench query compute should use new vec to make it compute for each query

What changes are included in this PR?

The average time for clickbench query compute should use new vec to make it compute for each query

Are these changes tested?

Yes

Q42: SELECT DATE_TRUNC('minute', to_timestamp_seconds("EventTime")) AS M, COUNT(*) AS PageViews FROM hits WHERE "CounterID" = 62 AND "EventDate"::INT::DATE >= '2013-07-14' AND "EventDate"::INT::DATE <= '2013-07-15' AND "IsRefresh" = 0 AND "DontCountHits" = 0 GROUP BY DATE_TRUNC('minute', to_timestamp_seconds("EventTime")) ORDER BY DATE_TRUNC('minute', M) LIMIT 10 OFFSET 1000;
Query 42 iteration 0 took 56.9 ms and returned 10 rows
Query 42 iteration 1 took 54.6 ms and returned 10 rows
Query 42 iteration 2 took 56.5 ms and returned 10 rows
Query 42 iteration 3 took 54.4 ms and returned 10 rows
Query 42 iteration 4 took 50.4 ms and returned 10 rows
Query 42 avg time: 54.57 ms

Before this PR:

Q42: SELECT DATE_TRUNC('minute', to_timestamp_seconds("EventTime")) AS M, COUNT(*) AS PageViews FROM hits WHERE "CounterID" = 62 AND "EventDate"::INT::DATE >= '2013-07-14' AND "EventDate"::INT::DATE <= '2013-07-15' AND "IsRefresh" = 0 AND "DontCountHits" = 0 GROUP BY DATE_TRUNC('minute', to_timestamp_seconds("EventTime")) ORDER BY DATE_TRUNC('minute', M) LIMIT 10 OFFSET 1000;
Query 42 iteration 0 took 52.7 ms and returned 10 rows
Query 42 iteration 1 took 48.1 ms and returned 10 rows
Query 42 iteration 2 took 53.8 ms and returned 10 rows
Query 42 iteration 3 took 52.8 ms and returned 10 rows
Query 42 iteration 4 took 50.2 ms and returned 10 rows
Query 42 avg time: 1121.93 ms

Are there any user-facing changes?

Fix the wrong average time.

[zhuqi-lucas]

cc @2010YOUY01 @xudong963

[xudong963]

Thank you