Aggregation query's shouldn't return rows when there isn't any data

We noticed that there is inconsistency in the way aggregation function queries return data.
For example
```
select sum(count_field) from table where application='xxxx'
```
This will return **1 row with value 0,** even when I make up a application value that doesn't exist.
If however, I added a time filter to this,
```
select created_15min, sum(count_field) from table where application = 'xxxx' and created_15min >= 1664582400000
group by 1
```
this returns **0 rows.**

I believe the correct response here should be zero rows, especially since for that first query for example, `numDocsScanned` is 0

Start by reproducing the two aggregation queries from the issue against Apache Pinot, comparing results when the filter matches no data and checking the reported numDocsScanned value. Trace the aggregation query path to identify where the no-data result differs, then add regression coverage showing the expected empty result and verify both query forms.

Written by the indexing model from the issue text.