On version 5.7, repeated results were obtained when using group by

The table creation statement is as follows：

The data in the table is as follows：

The table query statement is as follows：

The query results are as follows:

The execution steps of this query are as follows:

1. join_init_dead_decord: Sort the data from the privacey_apply table by the field corresponding to group by

2. rr_unpack_from-buffer, loop to retrieve a record from the sort buffer

3. evaluate_join_record -> sub_select_op -> QEP_tmp_table:: put_record -> end_write_group， Check if the current record is the same as the previous record. If it is different, it means it is a new group, and start a new count accordingly.
   
   The reason for this bug is that before the first step of sorting is performed, the value of the column corresponding to 'group by' has not been generated yet (generated in step 3) when Sort_param:: make_stortkey generates the sorting key, resulting in invalid sorting and causing a duplicate 'group' to be generated when executing end_write_group in step 3