Description:
From MySQL 8.0.20 and later version, I can see regression on a select group by sentences would not work correctly when specifying a timezone, the result returns duplicate dates.
Previous releases work fine.
How to repeat:
Prepare environment:
Apply attached SQL file.
Exec query, it's return expected output:
```
mysql> set time_zone = 'UTC';
mysql> select date(from_unixtime(date_received)) as `dt`, count(distinct pid) FROM test2 GROUP by `dt`;
+------------+---------------------+
| dt | count(distinct pid) |
+------------+---------------------+
| 2020-07-30 | 10 |
| 2020-07-31 | 10 |
| 2020-08-01 | 10 |
| 2020-08-02 | 10 |
+------------+---------------------+
4 rows in set (0.00 sec)
```
But when we change timezone and exec this query it's return output with duplicate dates:
```
mysql> set time_zone = 'America/Los_Angeles';
mysql> select date(from_unixtime(date_received)) as `dt`, count(distinct pid) FROM test2 GROUP by `dt`;
+------------+---------------------+
| dt | count(distinct pid) |
+------------+---------------------+
| 2020-07-30 | 3 |
| 2020-07-29 | 2 |
| 2020-07-30 | 2 |
| 2020-07-29 | 1 |
| 2020-07-30 | 1 |
| 2020-07-29 | 1 |
| 2020-07-30 | 1 |
| 2020-07-29 | 1 |
| 2020-07-30 | 1 |
| 2020-07-29 | 2 |
| 2020-07-30 | 7 |
| 2020-07-29 | 3 |
| 2020-07-30 | 2 |
| 2020-07-29 | 10 |
| 2020-07-30 | 1 |
| 2020-07-29 | 6 |
| 2020-07-30 | 1 |
| 2020-07-29 | 10 |
| 2020-07-30 | 10 |
| 2020-07-31 | 2 |
| 2020-07-30 | 1 |
| 2020-07-31 | 1 |
| 2020-07-30 | 1 |
| 2020-07-31 | 2 |
| 2020-07-30 | 2 |
| 2020-07-31 | 2 |
| 2020-07-30 | 2 |
| 2020-07-31 | 1 |
| 2020-07-30 | 1 |
| 2020-07-31 | 2 |
| 2020-07-30 | 1 |
| 2020-07-31 | 2 |
| 2020-07-30 | 10 |
| 2020-07-31 | 10 |
| 2020-08-01 | 4 |
| 2020-07-31 | 2 |
| 2020-08-01 | 1 |
| 2020-07-31 | 1 |
| 2020-08-01 | 5 |
| 2020-07-31 | 1 |
| 2020-08-01 | 1 |
| 2020-07-31 | 2 |
| 2020-08-01 | 2 |
| 2020-07-31 | 10 |
| 2020-08-01 | 1 |
| 2020-07-31 | 10 |
| 2020-08-01 | 10 |
+------------+---------------------+
47 rows in set (0.01 sec)
```
