Support Grouping for Varchar columns for sharded keyspaces

[GuptaManan100]

Feature Description

While ordering for all columns is supported after #7678, GROUP BY will not work for varchar columns since for now orderedAggregate still uses the original column for merging. i.e

# scatter group by a text column, reuse existing weight_string
"select count(*) k, a, textcol1, b from user group by a, textcol1, b order by k, textcol1"
{
  "QueryType": "SELECT",
  "Original": "select count(*) k, a, textcol1, b from user group by a, textcol1, b order by k, textcol1",
  "Instructions": {
    "OperatorType": "Sort",
    "Variant": "Memory",
    "OrderBy": "0 ASC, 2 ASC",
    "Inputs": [
      {
        "OperatorType": "Aggregate",
        "Variant": "Ordered",
        "Aggregates": "count(0)",
        "Distinct": "false",
        "GroupBy": "1, 4, 3",
        "Inputs": [
          {
            "OperatorType": "Route",
            "Variant": "SelectScatter",
            "Keyspace": {
              "Name": "user",
              "Sharded": true
            },
            "FieldQuery": "select count(*) as k, a, textcol1, b, weight_string(textcol1), weight_string(a), weight_string(b) from `user` where 1 != 1 group by a, textcol1, b",
            "OrderBy": "2 ASC, 1 ASC, 3 ASC",
            "Query": "select count(*) as k, a, textcol1, b, weight_string(textcol1), weight_string(a), weight_string(b) from `user` group by a, textcol1, b order by textcol1 asc, a asc, b asc",
            "Table": "`user`"
          }
        ]
      }
    ]
  }
}

Notice that in the plan output the ordered aggregate is using 1,4 and 3 columns instead of 5,4 and 6. To support this the orderedAggregate will also have to store both the column and the weight_string column.

Use Case(s)

GROUP BY on varchar columns in shared keyspace for tables which are not authoritative.

[harshit-gangal]

Support added with #8453