Question: aggregateWindow function doesn't work as my expectation

Nobody has claimed this yet.

Data Set:
|_start|_stop|_time|_value|_field|_measurementservice|
|--|--|--|--|--|--|
|2022-08-04 13:34:00 |2022-08-04 13:37:00 | 2022-08-04 13:34:00| 66.06 |CpuUsage |Mongo |prod|
|2022-08-04 13:34:00|2022-08-04 13:37:00|2022-08-04 13:35:00|68.38|CpuUsage|Mongoprod|
|2022-08-04 13:34:00|2022-08-04 13:37:00|2022-08-04 13:36:00|90.55|CpuUsage|Mongoprod|

Query:
```
from(bucket: "default")
|> range(start: 2022-08-04T05:34:00.000Z, stop: 2022-08-04T05:37:00.000Z)
|> filter(fn: (r) => r["_measurement"] == "Mongo")
|> filter(fn: (r) => r["_field"] == "CpuUsage" and r["service"] == "prod")
|> aggregateWindow(every: 3m, fn: mean, createEmpty: false)
```

Query Result:
|_start|_stop|_time|_value|_field|_measurement|service|
|--|--|--|--|--|--|--|
|2022-08-04 13:34:00|2022-08-04 13:37:00|2022-08-04 13:36:00|67.22|CpuUsage|Mongo|prod|
|2022-08-04 13:34:00|2022-08-04 13:37:00|2022-08-04 13:37:00|90.55|CpuUsage|Mongo|prod|

My Question:

1. With the query string above, I expected one line data, but actual result is two lines.
1. And I want to know the detail about how the func `aggregateWindow` works, I can't get more infomation from influxdb's doc.
1. how the parameter every of `aggregateWindow` split data to windows

Start with the aggregateWindow query and result shown in the issue, then review the existing aggregateWindow documentation and its description of window boundaries and the every parameter. Document why this query produces two rows and explain how every divides the range. Done means the behavior and parameter semantics are clear to users.

Written by the indexing model from the issue text.