Description
See screenshot for the following Druid query:
{code:java}
SELECT 
DATE_TRUNC('day', __time) AS time_date,
TIME_EXTRACT(__time, 'day') AS day_of_month,
SUM(quantity) AS total_quantity
FROM relay_billing_reporting
WHERE
__time > '2024-06-15'
AND productId = 'RLAY-CCU'
AND (TIME_EXTRACT(__time, 'day') >= 28 OR TIME_EXTRACT(__time, 'day') <= 2)
GROUP BY TIME_EXTRACT(__time, 'day'), DATE_TRUNC('day', __time)
ORDER BY time_date {code}
!image-2025-02-20-13-50-57-080.png|width=910,height=884!

Reproduction steps

Visit Duid UI to inspect the data:
[https://imply.prd-imply.unity3d.com/e77ce717-dafa-4623-90ac-4eea158eea80/unified-console.html#query]

Select relay_billing_reporting datasource and type the query above.

Expected behaviour
The time series should be continuous on a day by day basis.

Actual behavior
There's a significant spike on the first day of every month.