…able with random distribute (apache#52993)

If agg table is random hash distribute, would add aggregate node on
scan.
The aggregate function alias expr id is same to the child expr id of
alias.

such as query sql is `select * from db1.tagg`
the query plan is as following, and the `sum(b#1) AS `b`#1`, alias expr
id is same to the child expr id of alias, the id is 1
this would cause hidden problems
```sql
LogicalResultSink[30] ( outputExprs=[a#0, b#1] )
+--LogicalAggregate[29] ( groupByExpr=[a#0], outputExpr=[a#0, sum(b#1) AS `b`#1], hasRepeat=false, stats=1 )
   +--LogicalOlapScan ( qualified=internal.db1.tagg, indexName=<index_not_selected>, selectedIndexId=1752042452160, preAgg=ON, operativeCol=[a#0, b#1], stats=1 )
```

the pr fix this, and the expression  change to `sum(b#1) AS `b`apache#4`  
```sql
LogicalResultSink[30] ( outputExprs=[a#0, b#4] )
+--LogicalAggregate[29] ( groupByExpr=[a#0], outputExpr=[a#0, sum(b#1) AS `b`apache#4], hasRepeat=false, stats=1 )
   +--LogicalOlapScan ( qualified=internal.db1.tagg, indexName=<index_not_selected>, selectedIndexId=1752042065062, preAgg=ON, operativeCol=[a#0, b#1], stats=1 )
```